_0_1_mlxtend.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:38:31.027974Z",
     "start_time": "2018-04-19T09:38:27.753051Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/duoan/.pyenv/versions/stats36/lib/python3.6/site-packages/sklearn/preprocessing/label.py:151: DeprecationWarning: The truth value of an empty array is ambiguous. Returning False, but in future this will result in an error. Use `array.size > 0` to check that an array is not empty.\n",
      "  if diff:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x110b74dd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.gridspec as gridspec\n",
    "import itertools\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "\n",
    "# Initializing Classifiers\n",
    "clf1 = LogisticRegression(random_state=0)\n",
    "clf2 = RandomForestClassifier(random_state=0)\n",
    "clf3 = SVC(random_state=0, probability=True)\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1, clf2, clf3],\n",
    "                              weights=[2, 1, 1], voting='soft')\n",
    "\n",
    "# Loading some example data\n",
    "X, y = iris_data()\n",
    "X = X[:,[0, 2]]\n",
    "\n",
    "# Plotting Decision Regions\n",
    "\n",
    "gs = gridspec.GridSpec(2, 2)\n",
    "fig = plt.figure(figsize=(10, 8))\n",
    "\n",
    "labels = ['Logistic Regression',\n",
    "          'Random Forest',\n",
    "          'RBF kernel SVM',\n",
    "          'Ensemble']\n",
    "\n",
    "for clf, lab, grd in zip([clf1, clf2, clf3, eclf],\n",
    "                         labels,\n",
    "                         itertools.product([0, 1],\n",
    "                         repeat=2)):\n",
    "    clf.fit(X, y)\n",
    "    ax = plt.subplot(gs[grd[0], grd[1]])\n",
    "    fig = plot_decision_regions(X=X, y=y,\n",
    "                                clf=clf, legend=2)\n",
    "    plt.title(lab)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Adaptive Linear Neuron -- Adaline\n",
    "## Example 1 - Closed Form Solution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:40:07.438066Z",
     "start_time": "2018-04-19T09:40:07.203117Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/duoan/.pyenv/versions/stats36/lib/python3.6/site-packages/mlxtend/_base/_classifier.py:19: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
      "  if not np.issubdtype(y[0], int):\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109b4c9e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import Adaline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "\n",
    "ada = Adaline(epochs=30, \n",
    "              eta=0.01, \n",
    "              minibatches=None, \n",
    "              random_seed=1)\n",
    "ada.fit(X, y)\n",
    "plot_decision_regions(X, y, clf=ada)\n",
    "plt.title('Adaline - Stochastic Gradient Descent')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:40:28.132409Z",
     "start_time": "2018-04-19T09:40:27.570717Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/duoan/.pyenv/versions/stats36/lib/python3.6/site-packages/mlxtend/_base/_classifier.py:19: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
      "  if not np.issubdtype(y[0], int):\n",
      "Iteration: 30/30 | Cost 3.79 | Elapsed: 0:00:00 | ETA: 0:00:0000"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAEICAYAAABLdt/UAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAHH1JREFUeJzt3XuUXGWZ7/Hv092VkJALSSchpMPdDBdRYIY7I+NRGFtWBJGJQwQ0g0w4RMcxygFGWTOwxBk5DgSPOnOSJYgXJOOMBiFoEjygiAMGZKImgQAikHRuE0jShEDoy3P+2LtJ0anu2l311t5Vu3+ftXp1Ve1d7/vsXd1Pv/Xup98yd0dERPKjKesAREQkLCV2EZGcUWIXEckZJXYRkZxRYhcRyRkldhGRnFFiz5CZzTGzhxPue72ZfTe+fYiZ7TKz5tpGmB0zu8PMbkyhn4vNbEWt+6lU8Xkws3eZ2bqsY5L6p8ReA2b2MzPbbmYja9G+u7/o7mPcvacW7Q/GzKab2Q/MbJuZ7TSz1WY2J952mJm5mbWkHVcSpeJz9zvd/c8rbO8kM1sav9Y7zGytmX3RzCaEi3ovd/+Fux8Voi0ze97Mzh5k+7vNrDceQOwysw1m9n0zOzlE/7UQ/95dnnUc9UCJPTAzOwx4F+DAeZkGUxvfAdYDhwKtwKXAlkwjyoCZnQH8DPglcLS7HwC0A93A8QM8py7/4A1io7uPAcYCpwFPAb8ws/dmG5aU5e76CvgF/D3RL/stwNJ+21qBe4BOYCXwBeDhou1fIUqancCvgXcVbbse+G58+zCiPxwt8f2fxW39EngFWAFMKnruacB/AjuA3wDvruL4dgEnDLDtxTiuXfHX6USDh+uAF4CtwLeB8UXP+dOi2NYDc+LH7wC+DtwXH9OvgCMTnqtTgMfjbVuAWwaJb06/1+DtwP3Ay/FzPzfAsT4MfLXMuZoTvyYLgJeAG4EjgQfi+9uAO4EDip5zIvBEfMz/BiwGboy3vRvYULTvNOAHwH8DfwA+1e/n5fvx+X4FWAOcFG/7DtALvBafh6tLxP6Wvooe/xrweNH9o4vO1zrgw0XbzgXWxv13AFcVbTsfWBW/Rr8H2uPHxwO3AZvi59wINBedz4eBfwa2x8f8/njbF4Ee4PX4mL6WdS7I8ivzAPL2BTwLzAP+BOgCDizatjj+ZdsfOC7+wS1OKpcQJf8W4LPAZmC/eNv1DJ7Yfw/8ETAqvv+leFtbnETOJUqy58T3J1d4fD8lSlYXAYf02/aWuOLHLovPyRHAGOCHwHfibYfGv/SzgUJ87CfE2+6I4zwlPh93AosTnqtHgEvj22OA0waJb07fa0A0Mt0Ut7dffP/UEudg/ziJvLvMuZpDNIL/mzjOUcDb4tdgJDAZeAi4Nd5/BNEfwPnx+fiL+Gdon8Qev5a/JhpIjIjP73PA+4p+Xl6PX/dm4J+AR4tiex44e5DY3+yr3+PvIfqjsH/8tR74q/j4TiT6Y3VsvO8m4j+4wATgj+PbpwA74/PQRPQzenS8bQmwMG57CtEA6Iqi89kF/HV8TFcCGwEr+j24POscUA9fmQeQpy+i0WcX8WiZ6K3r/Ph2c7zt6KL9/5GixF6ive3A8fHt6xk8sV9X9Lx5wLL49jXEibRo+3LgYxUe4wTgS0QjwB6iUdfJpeKKH/t/wLyi+0fF56EF+DtgyQD93AF8o+j+ucBTCc/VQ8ANFL1rGSS+OexN7LOB/0pwDqbH7RS/lv+b6F3Hq32vRdz2i2Xa+mBfn8BZxYkqfuw/KZ3YT+3fdnw+v1n08/LTom3HAq8V3X+eyhL70fGxtwF/Cfyi3/aFwD/Et18ErgDGldhnQYm2DwT2AKOKHpsNPFh0Pp8t2jY6jmVq0e+BEru75tgD+xiwwt23xfe/Fz8G0eishWiE0+eF4ieb2VVm9mR8UXIH0dvSSQn73lx0ezfRSBWiUfGs+OLejrjdPwUO6t9AXCHSd7HsJ6U6cfft7n6tu7+d6BdxFXC3mdkAcU3rd5wvEJ2HA4GDid5pDPWYyp2rjxO9e3nKzB4zs5mD9FGsXDx9thONWt88h+5+tUfz7EuIjq9P8euNmR1oZovNrMPMOoHvFsU9DejwOEvF3vIzUuRQYFq/1/VzROe1T//zt1+Aef42omS6I47h1H4xXAxMjfe9kOgP8gtm9nMzOz1+fKDzfCjRO5VNRe0tJBq573NM7r47vjkGeYtGu5hTt8xsFPBhoNnM+n74RgIHmNnxwGqit+UHE43kAQ4pev67gKuB9wJr3L3XzLYDAyXMpNYTjdj/utyO7n4n0ZRHIu6+zcz+meiP10SiX/j+NhL9wvY5hOg8bIljOyVpf33KnSt3fwaYbWZNwIeA/zCz1gHiK7aeaIppUO7+qpn9Km77wXK797v/j/Fj73D3l83sg0Tz1hBNXbSZmRUl90MonQTXA39w9xnl4k0YV1IXAE/E52A98HN3P6dkB+6PAeebWQH4JNE05MFEsR9Z4inriUbsk9y9u4LYKj2m3NGIPZwPEk1NHAucEH8dA/wC+KhHpYk/BK43s9Fmdix7R/MQzed2E10IazGzvwfGBYjru8AHzOx9ZtZsZvvFpWzTK2nMzG4ys+PMrMXMxhLNcz7r7i/FsfcSzff2uQuYb2aHm9kYosT2b/Ev7p3A2Wb24bi9VjM7IUEYg54rM7vEzCa7ey/RyJI4rlLxFVsKHGRmnzazkWY21sxOHWDfq4HLzOxaM5sS9zsdODxB7LuAnWbWBvyvom2PxMf1KTMrmNmHGPgP30rgFTO7xsxGxa/tcUMoR9zCwOfhLSzSZmb/AFxO9M4AovP1R2Z2aRxvwcxONrNjzGxE/A5wvLt3EV0k7Y2fdxvwV2b2XjNrits+2t03EV34v9nMxsXbjjSzPwt9THmnxB7Ox4jmN1909819X0SjsYvjt8CfJHrbuJloDvmbRc9fDiwDniZ6+/06/d7GV8Ld1xNVIHyOKLGtJ0omlb72o4mmG3YQXaw7lLisM35r/EXgl/Fb6dOA24mqMB4iqmJ4nehiIu7+ItFb9c8SVVWsYoBSwX7Knat2YI2Z7SKqnrnI3V8bIL43ufsrRBf0PkD0Gj0D/I9SAbj7w0QXEs8Cno6nDZYRzfN+dZDYbwD+mOji4X1Ef+z72nyD6F3AHKLz8ZfF2/v13wPMJBpA/IHoouU3iKakkvgn4Lr4PFw1wD7T4nO4C3gMeAfRBeMVcQyvAH9O9C5nI9E5u4nonSpEpbDPx1NO/5NomgZ3X0l0wXVBfB5+zt53dR8luhi8lmjK6z8oMW04gK8Af2HR/xX8n4TPySVz17sXEZE80YhdRCRnlNhFRHJGiV1EJGeU2EVEciaTOvYnnlC9qYjIUIwdCzNmJPu/lkwSe0dHFr2KiDSu1laYkfDf0TQVIyKSM0rsIiI5o8QuIpIzWgRMRIaRXkaPfpVCIfVPlRySrq5mdu/en0rH3krsIjJsjB79KmPGFIjWo6t24dRacUaO3AO8yu7dYytqQVMxIjJsFAo9RJ8xX69JHcAwG1nVuwoldhEZZuo5qfepLkYldhGRnFFiFxFJ2SOPLGPWrKO48MK38a1vfSl4+0rsIiIp6unp4ctf/gS33voTFi9ey4oVd/Hcc2uD9qGqGBGREv5mzsns3r51n8dHT5jCV+94rOJ2165dyfTpb6OtLfoUv3POuYiHHvoRRxxxbMVt9qfELiJSwu7tW3n0gMn7PH5aiWQ/FFu3dnDggQe/eX/KlOmsWfOrqtrsT1MxIiI5o8QuIpKiKVPa2LJl72evb926gcmT24L2ocQuIpKiY445mfXrn2Hjxj/Q1fUG99+/mLPOOi9oH5pjFxFJUUtLC1dd9TU+9an30dvbwwc+cBlHHPH2sH0EbU1EJCdGT5hS8kLp6AlTqm77zDPP5cwzz626nYEosYuIlFBNSWPWNMcuIpIzSuwiIjmjxC4ikjNK7CIiOaPELiKSM0rsIiIp+8IXLqO9fQqzZx9Xk/aV2EVEUjZz5hxuvXVZzdpXYhcRGcSOHduYP/9Cdu58KVibJ554FuPGTQzWXn9K7CIig7j77m+zZs12liz5VtahJKbELiIygB07tnHffUuZOPFfue++pUFH7bWkxC4iMoC77/42PT0zGTnyKHp6ZjbMqF2JXUSkhL7R+qhRHwVg1KiPNsyoverEbmYHm9mDZrbWzNaY2d+GCExEJEt9o/WWlkkAtLRMCjZqv+662Vx++em88MI6Zs6czj333FZ1m8VCrO7YDXzW3Z8ws7HAr83sfncP+7HbIiIpWrnyZ/T0bGTHju/1e3wac+Z8pqq2b7zxrqqeX07Vid3dNwGb4tuvmNmTQBugxC4iDetf/uWerEOoWNA5djM7DDgR2Ocjt81srpk9bmaPL1u2KGS3IiJSJNgHbZjZGOAHwKfdvbP/dndfBCwCuPdePFS/IiJD44BlHUQZ1aXIICN2MysQJfU73f2HIdoUEQmtq6sZ9z1Umzhry3HfQ1dXc8UtVD1iNzMDbgOedPdbqm1PRKRWdu/eH3iVQuH1rEMZVFdXcxxrZUJMxZwJXAr8zsxWxY99zt1/HKBtEZGAmti9e2zWQdRciKqYh6n/CSsRkWFD/3kqUkOdndu44YYL6eys//9WlPxQYhepoeXLv83TT29n+fLGWGNE8kGJXaRGOju38cADS5kw4V954IGlGrVLapTYRWpk+fJorZERI6KVATVql7QosYvUQN9ovXhlQI3aJS1K7CI10Ddab26OVgZsbp6kUbukJtiSAiKy16pV0cqAO3d+r9/j05g1q7qVAUXKMff0/7VWa8WIDE+dndtYsOAK5s9fxLhxrVmH01BaW+GMM5L9z5CmYkQkNSr/TIcSu4ikQuWf6VFiF5FUqPwzPUrsIlJzKv9MlxK7iNScyj/TpXJHEak5lX+mS+WOIiINQOWOIg2ko2Mdl1xyOB0dz9SsDy0fPLwosYtkbOHCa+nsnMjChVfXrA/Vjw8vSuwiGeroWMfq1b+jUPgmq1f/riajdtWPDz9K7CIZWrjwWnp7P0JT0zvp7f1ITUbtqh8ffpTYRTLSN1pvbr4SgObmK4OP2lU/PjwpsYtkZO9o/SAAmpoOCj5qV/348KQ6dpGMPPPMfwEr6eq6rd/jhWB9qH58eFIdu4hIA1Adu+ReiLrsNGq7k/SRxrFUuz2tONNqI++U2KUhhajLTqO2O0kfaRxLtdvTijOtNvJOiV0aToi67DRqu5P0kcaxVLs9rTjTamM4UGKXhhOiLjuN2u4kfaRxLNVuTyvOtNoYDpTYpaGEqMtOo7Y7SR9pHEu129OKM602hgsldmkoIeqy06jtTtJHGsdS7fa04kyrjeFCdezSUELUZadR252kjzSOpdrtacWZVhvDherYJXWdndtYsOAK5s9fxLhxrVmHU/c6OtZxzTXt3HTTCtraZmQdzoCSvK7l9gnxs5HXny/VsUtdU7na0KSxrG8IIUomVQ4ZhhK7pErlakOTxrK+IYQomVQ5ZDhK7JIqlasNTRrL+oYQomRS5ZDhBEnsZna7mW01s9Uh2pN8Urna0KSxrG8IIUomVQ4ZVqgR+x1Ae6C2JKdUrjY0aSzrG0KIkkmVQ4YVpNzR3R8ys8NCtCX5pXK1oUljWd8QQpRMqhwyrGDljnFiX+ruxw2wfS4wF2DevIV/0t4+N0i/IiLDQV2WO7r7Inc/yd1PUlKXNHR0rOOSSw4fcE563bpHueCCiaxbt3LANpLsU2shltQNsWxviDglHaqKkdwqV/99881X0tNzGDfffMWAbSTZp9bSqA9X/Xi+KLFLLpWr/1637lE2b94O3M7mzdtLjsiT7FNradSHq348f0KVO94FPAIcZWYbzOzjIdoVqVS5+u+bb74SuBizE4CLS47Ik+xTa2nUh6t+PH+CJHZ3n+3uB7l7wd2nu/tt5Z8lUhvl6r/3jsQ/ET/jE/uMyJPsU2tp1IerfjyfNBUjuVOu/nvvSHwaQPz9rSPyJPvUWhr14aofzyct2yu5U67+e+vW9cBt9H9juXVrd9Ht8vvUWhr14aofzyct2ysi0gDqso5dJG1p1GaXq5UPQfXhMlRK7JJbadRmp7FWuurDZaiU2CWX0qjNTmOtdNWHSyWU2CWX0qjNTmOtdNWHSyWU2CV30qjNTmOtdNWHS6WU2CV30qjNTmOtdNWHS6VUxy65k0Ztdhprpas+XCqlOnYRkQagOnbJVL3UXaexBrlIPVJil+Dqpe46jTXIReqRErsEVS9112msQS5Sr5TYJah6qbtOYw1ykXqlxC7B1EvddRprkIvUMyV2CaZe6q7TWINcpJ6pjl2CqZe66zTWIBepZ6pjl7fo7NzGggVXMH/+IsaNay25T0fHOq65pp2bblpBW9uMlCNMLsmxVNtGGn2IgOrYpQpJSgDTWKo2hDSW7U2jD5GhUmKXNyUpAUxjqdoQ0li2N40+RCqhxC5vSlICmMZStSGksWxvGn2IVEKJXYBkJYBpLFUbQhrL9qbRh0illNgFSFYCmMZStSGksWxvGn2IVErljgIkK1VMY6naENJYtjeNPkQqpXJHEZEGMJRyR43YRUQyNrWwjZG8Meg+Y5oLwORE7Smxi4gENKGlk8PthcT7F/wNpnRtLL9jTytK7CIiJTTRy35Ng4+Oi41seoPj/TcUPNlzRnS/ThO9lYYXhBK7iDS8aYX/ppmeZPt6B1O6E4yQ+2SboyuixC4ideXAlpeYyubE+49mN5O6ku8/HCixi8iQNNHLiKbuRPsWrJvj7TeM6N2TuP3R3bswFc5VRYldZJib0vIyLSRL1GZwRO+zHNCj/46tZ0rsIjkzvbCFCWxPtO9+/hpTuzfUOCJJmxK7SIoMp8WSXeQD2L/pNd7Jb4dUZTG2a0cloUmOBEnsZtYOfAVoBr7h7l8K0a5IvTOcKS0vY5ZsTvgwf57J3ZuSd5D8b4DIm6pO7GbWDHwdOAfYADxmZve4+9pq2xZJ24EtLzHJks8fj/XOoSVqkRSEGLGfAjzr7s8BmNli4HxAiV2CM3xI0xITWzo52p+EBGsimcG47u2qyJCGFyKxtwHri+5vAE7tv5OZzQXmAsybt5D29rkBupZGN6llB+bJErUZHONrGdeT7MIgQMJiD5FcSe3iqbsvAhaBVnfMs+mFLYxhV6J9x7CLg7perHFEIsNPiMTeARxcdH96/JjUq4RLNY9veZW329rEUxNN3sMBXapvFslaiMT+GDDDzA4nSugXAR8J0K6UYTgTWzoTJ+pm6+WdvasY1ftqsg5UkSHSkKpO7O7ebWafBJYTlTve7u5rqo5sGJpa2MY4Xkm8/0R/SRUZIrKPIHPs7v5j4Mch2sqTKS0vcyS/T7Sv4Uzs2qqKDBGp2rD+z9NmehjXkmxaognnHf5b9u/embj9pm4laRFJX64Se1thK6N4LfH+U30jE7q3Da2TRJ84KCKSnbpN7BNaOpnBs4kvDLZYD61dW2oclYhI/csksb/flpXdp7m7W/PNIiIVyCSxt3hXFt2KiAwLdTsVI/XnPZ/5DJ079714PG78eB645ZYMIhKRUpTYJbHOnTt5fPz4fR4/qUSyF5HsNGUdgIiIhKXELiKSM0rsIiI5o8QuIpIzungqiY0bP77khdJxJS6oikh2lNglMZU0ijQGJXYJRnXuIvVBiV2CUZ27SH3QxVMRkZxRYhcRyRkldhGRnFFiFxHJGV08lWBU5y5SH5TYJbFy5Yzr1q+Hrn3X2t/U2Rkshtbzz6dQ4lO1usx46Uc/CtaPSjelkSmxS2Jlyxm7uugoFPbZ3lYi2Veq4M5m2/eDZ6cm/AjFpFS6KY1Mc+wiIjmjxC4ikjNK7CIiOaPELiKSM7p4KomVLWcsFEpfKC1xQbVSXWYlL5R2lbigWg2VbkojMw9cTZDIvfdm0KmISANrbYUzzkg0gtGIvQ6Uq5kOUVOdRhtts2aVrGOnUKDj3/89UR/1QnXs0siU2OtAuZrpEDXVqbSRQh17WlTHLo1MF09FRHJGiV1EJGeU2EVEckaJXUQkZ6q6eGpms4DrgWOAU9z98RBBDTflaqZD1FSn0kYKdexpUR27NLKq6tjN7BigF1gIXJU4sauOXURkaNKqY3f3JwEs8H/95UmI9cPLtZGkfrxcG0nqtiecdx4jS8S3B9h+zz1B4gxRK59GPb3q3KWepVbHbmZzgbkAC+fNY257e1pdZyrE+uFl20hQP16ujSR12yOBzSXimxowziC18inU06vOXepZ2cRuZj9l7+9usc+7e+KPrHH3RcAiQFMxIiI1VDaxu/vZaQQiIiJhqNxRRCRnqkrsZnaBmW0ATgfuM7PlYcISEZFKVVsVswRYEiiWXAqxfnjZNhLUj5drI0nd9h5KX2zZEzDOILXyKdTTq85d6pnWY69So5S9lYszRFlmtTGIyCC0Hnt6GqXsrVycIcoyq41BRMLQxVMRkZxRYhcRyRkldhGRnFFiFxHJGV08rVKjlL2VizNEWWa1MYhIGCp3FBFpBEMod9RUjIhIziixi4jkjBK7iEjOKLGLiOSMEruISM4osYuI5IwSu4hIziixi4jkjBK7iEjOKLGLiOSMEruISM4osYuI5IwSu4hIziixi4jkjBK7iEjOKLGLiOSMEruISM4osYuI5IwSu4hIziixi4jkjBK7iEjOKLGLiOSMEruISM4osYuI5IwSu4hIziixi4jkTFWJ3cy+bGZPmdlvzWyJmR0QKjAREalMtSP2+4Hj3P2dwNPA31UfkoiIVKOqxO7uK9y9O777KDC9+pBERKQaIefYLwN+MtBGM5trZo+b2eOLli0L2K2IiBQzdx98B7OfAlNLbPq8u/8o3ufzwEnAh7xcgwD33lt+HxER2au1Fc44w5Ls2lJuB3c/e7DtZjYHmAm8N1FSFxGRmiqb2AdjZu3A1cCfufvuMCGJiEg1qp1j/xowFrjfzFaZ2f8NEJOIiFShqhG7u78tVCAiIhKG/vNURCRnlNhFRHJGiV1EJGeqmmOv2PjxmXQrItKwxoxJvGvZf1CSiJnNdfdFWccRWh6PK4/HBDquRpPlcWkqJrm5WQdQI3k8rjweE+i4Gk1mx6XELiKSM0rsIiI5o8SeXO7mAGN5PK48HhPouBpNZseli6ciIjmjEbuISM4osYuI5IwS+xDk8cO7zWyWma0xs14zOynreKplZu1mts7MnjWza7OOJwQzu93MtprZ6qxjCcXMDjazB81sbfzz97dZxxSCme1nZivN7Dfxcd2QRRxK7EOTxw/vXg18CHgo60CqZWbNwNeB9wPHArPN7NhsowriDqA96yAC6wY+6+7HAqcBn8jJa7UHeI+7Hw+cALSb2WlpB6HEPgR5/PBud3/S3ddlHUcgpwDPuvtz7v4GsBg4P+OYqubuDwEvZx1HSO6+yd2fiG+/AjwJtGUbVfU8siu+W4i/Uq9QUWKv3KAf3i2ZaAPWF93fQA6SRd6Z2WHAicCvso0kDDNrNrNVwFbgfndP/biyWQSsjg3hw7u7gTvTjK1SSY5JJAtmNgb4AfBpd+/MOp4Q3L0HOCG+BrfEzI5z91Svjyix95PHD+8ud0w50gEcXHR/evyY1CEzKxAl9Tvd/YdZxxOau+8wsweJro+kmtg1FTMERR/efZ4+vLsuPQbMMLPDzWwEcBFwT8YxSQlmZsBtwJPufkvW8YRiZpP7quXMbBRwDvBU2nEosQ9N7j6828wuMLMNwOnAfWa2POuYKhVf2P4ksJzoYtz33X1NtlFVz8zuAh4BjjKzDWb28axjCuBM4FLgPfHv0iozOzfroAI4CHjQzH5LNNC4392Xph2ElhQQEckZjdhFRHJGiV1EJGeU2EVEckaJXUQkZ5TYRURyRoldRCRnlNhFRHLm/wPQ1efpnXyYFAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109b4c0b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x110cb62e8>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109ba4668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import Adaline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "\n",
    "ada = Adaline(epochs=30, \n",
    "              eta=0.01, \n",
    "              minibatches=1, # for Gradient Descent Learning\n",
    "              random_seed=1,\n",
    "              print_progress=3)\n",
    "\n",
    "ada.fit(X, y)\n",
    "plot_decision_regions(X, y, clf=ada)\n",
    "plt.title('Adaline - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(ada.cost_)), ada.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 3 - Stochastic Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:41:09.810772Z",
     "start_time": "2018-04-19T09:41:09.271789Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/duoan/.pyenv/versions/stats36/lib/python3.6/site-packages/mlxtend/_base/_classifier.py:19: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
      "  if not np.issubdtype(y[0], int):\n",
      "Iteration: 15/15 | Cost 3.81 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109b28a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109b28048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import Adaline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "\n",
    "ada = Adaline(epochs=15, \n",
    "              eta=0.02, \n",
    "              minibatches=len(y), # for SGD learning \n",
    "              random_seed=1,\n",
    "              print_progress=3)\n",
    "\n",
    "ada.fit(X, y)\n",
    "plot_decision_regions(X, y, clf=ada)\n",
    "plt.title('Adaline - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(ada.cost_)), ada.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 4 - Stochastic Gradient Descent with Minibatches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:41:43.010582Z",
     "start_time": "2018-04-19T09:41:42.556064Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/duoan/.pyenv/versions/stats36/lib/python3.6/site-packages/mlxtend/_base/_classifier.py:19: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
      "  if not np.issubdtype(y[0], int):\n",
      "Iteration: 15/15 | Cost 3.87 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109b28be0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10abb1358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import Adaline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "\n",
    "ada = Adaline(epochs=15, \n",
    "              eta=0.02, \n",
    "              minibatches=5, # for SGD learning w. minibatch size 20\n",
    "              random_seed=1,\n",
    "              print_progress=3)\n",
    "\n",
    "ada.fit(X, y)\n",
    "plot_decision_regions(X, y, clf=ada)\n",
    "plt.title('Adaline - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(ada.cost_)), ada.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:42:11.990260Z",
     "start_time": "2018-04-19T09:42:11.986984Z"
    }
   },
   "source": [
    "# EnsembleVoteClassifier\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/EnsembleVoteClassifier_files/majority_voting.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 1 - Classifying Iris Flowers Using Different Classification Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:43:22.692889Z",
     "start_time": "2018-04-19T09:43:22.686344Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, 1:3], iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:43:52.111600Z",
     "start_time": "2018-04-19T09:43:52.011046Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5-fold cross validation:\n",
      "\n",
      "Accuracy: 0.90 (+/- 0.05) [Logistic Regression]\n",
      "Accuracy: 0.93 (+/- 0.05) [Random Forest]\n",
      "Accuracy: 0.91 (+/- 0.04) [Naive Bayes]\n"
     ]
    }
   ],
   "source": [
    "from sklearn import model_selection\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import numpy as np\n",
    "\n",
    "clf1 = LogisticRegression(random_state=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "\n",
    "print('5-fold cross validation:\\n')\n",
    "\n",
    "labels = ['Logistic Regression', 'Random Forest', 'Naive Bayes']\n",
    "\n",
    "for clf, label in zip([clf1, clf2, clf3], labels):\n",
    "    scores = model_selection.cross_val_score(clf, X, y, cv=5, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:48:31.751735Z",
     "start_time": "2018-04-19T09:48:31.748334Z"
    }
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings(action=\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:48:32.684187Z",
     "start_time": "2018-04-19T09:48:32.501474Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.90 (+/- 0.05) [Logistic Regression]\n",
      "Accuracy: 0.93 (+/- 0.05) [Random Forest]\n",
      "Accuracy: 0.91 (+/- 0.04) [Naive Bayes]\n",
      "Accuracy: 0.95 (+/- 0.05) [Ensemble]\n"
     ]
    }
   ],
   "source": [
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1, clf2, clf3], weights=[1, 1, 1])\n",
    "\n",
    "labels = ['Logistic Regression', 'Random Forest', 'Naive Bayes', 'Ensemble']\n",
    "for clf, label in zip([clf1, clf2, clf3, eclf], labels):\n",
    "    scores = model_selection.cross_val_score(clf, X, y, cv=5, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plotting Decision Regions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:49:12.075374Z",
     "start_time": "2018-04-19T09:49:07.499821Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112e4d668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "import matplotlib.gridspec as gridspec\n",
    "import itertools\n",
    "\n",
    "gs = gridspec.GridSpec(2, 2)\n",
    "fig = plt.figure(figsize=(10, 8))\n",
    "labels = ['Logistic Regression', 'Random Forest', 'Naive Bayes', 'Ensemble']\n",
    "for clf, lab, grd in zip([clf1, clf2, clf3, eclf], labels, itertools.product([0, 1], repeat=2)):\n",
    "    clf.fit(X, y)\n",
    "    ax = plt.subplot(gs[grd[0], grd[1]])\n",
    "    fig = plot_decision_regions(X=X, y=y, clf=clf)\n",
    "    plt.title(lab)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Grid Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:50:25.964419Z",
     "start_time": "2018-04-19T09:50:25.957319Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, 1:3], iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:50:40.860459Z",
     "start_time": "2018-04-19T09:50:37.701775Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.953 +/- 0.01 {'logisticregression__C': 1.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.960 +/- 0.01 {'logisticregression__C': 1.0, 'randomforestclassifier__n_estimators': 200}\n",
      "0.960 +/- 0.01 {'logisticregression__C': 100.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.953 +/- 0.02 {'logisticregression__C': 100.0, 'randomforestclassifier__n_estimators': 200}\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "\n",
    "clf1 = LogisticRegression(random_state=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1, clf2, clf3], voting='soft')\n",
    "\n",
    "params = {\n",
    "    'logisticregression__C': [1.0, 100.0],\n",
    "    'randomforestclassifier__n_estimators': [20, 200],\n",
    "}\n",
    "\n",
    "grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)\n",
    "grid.fit(iris.data, iris.target)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\" % (grid.cv_results_[cv_keys[0]][r],\n",
    "                                  grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "                                  grid.cv_results_[cv_keys[2]][r]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:52:23.392348Z",
     "start_time": "2018-04-19T09:52:17.980746Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.967 +/- 0.01 {'logisticregression-1__C': 1.0, 'logisticregression-2__C': 1.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.967 +/- 0.01 {'logisticregression-1__C': 1.0, 'logisticregression-2__C': 1.0, 'randomforestclassifier__n_estimators': 200}\n",
      "0.967 +/- 0.01 {'logisticregression-1__C': 1.0, 'logisticregression-2__C': 100.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.960 +/- 0.01 {'logisticregression-1__C': 1.0, 'logisticregression-2__C': 100.0, 'randomforestclassifier__n_estimators': 200}\n",
      "0.967 +/- 0.01 {'logisticregression-1__C': 100.0, 'logisticregression-2__C': 1.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.960 +/- 0.01 {'logisticregression-1__C': 100.0, 'logisticregression-2__C': 1.0, 'randomforestclassifier__n_estimators': 200}\n",
      "0.960 +/- 0.01 {'logisticregression-1__C': 100.0, 'logisticregression-2__C': 100.0, 'randomforestclassifier__n_estimators': 20}\n",
      "0.960 +/- 0.01 {'logisticregression-1__C': 100.0, 'logisticregression-2__C': 100.0, 'randomforestclassifier__n_estimators': 200}\n"
     ]
    }
   ],
   "source": [
    "clf1 = LogisticRegression(random_state=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1, clf1, clf2], voting='soft')\n",
    "\n",
    "params = {\n",
    "    'logisticregression-1__C': [1.0, 100.0],\n",
    "    'logisticregression-2__C': [1.0, 100.0],\n",
    "    'randomforestclassifier__n_estimators': [20, 200],\n",
    "}\n",
    "\n",
    "grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)\n",
    "grid = grid.fit(iris.data, iris.target)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\" % (grid.cv_results_[cv_keys[0]][r],\n",
    "                                  grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "                                  grid.cv_results_[cv_keys[2]][r]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 3 - Majority voting with classifiers trained on different feature subsets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Feature selection algorithms implemented in scikit-learn as well as the SequentialFeatureSelector implement a transform method that passes the reduced feature subset to the next item in a Pipeline.\n",
    "\n",
    "For example, the method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:53:35.354965Z",
     "start_time": "2018-04-19T09:53:35.351605Z"
    }
   },
   "outputs": [],
   "source": [
    "def transform(self, X):\n",
    "    return X[:, self.k_feature_idx_]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:54:30.196417Z",
     "start_time": "2018-04-19T09:54:20.572166Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.953 +/- 0.01 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 1, 'randomforestclassifier__n_estimators': 20}\n",
      "0.947 +/- 0.02 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 1, 'randomforestclassifier__n_estimators': 200}\n",
      "0.953 +/- 0.01 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 2, 'randomforestclassifier__n_estimators': 20}\n",
      "0.947 +/- 0.02 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 2, 'randomforestclassifier__n_estimators': 200}\n",
      "0.953 +/- 0.01 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 3, 'randomforestclassifier__n_estimators': 20}\n",
      "0.953 +/- 0.02 {'pipeline__logreg__C': 1.0, 'pipeline__sfs__k_features': 3, 'randomforestclassifier__n_estimators': 200}\n",
      "0.947 +/- 0.02 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 1, 'randomforestclassifier__n_estimators': 20}\n",
      "0.953 +/- 0.02 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 1, 'randomforestclassifier__n_estimators': 200}\n",
      "0.947 +/- 0.02 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 2, 'randomforestclassifier__n_estimators': 20}\n",
      "0.947 +/- 0.02 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 2, 'randomforestclassifier__n_estimators': 200}\n",
      "0.960 +/- 0.01 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 3, 'randomforestclassifier__n_estimators': 20}\n",
      "0.953 +/- 0.02 {'pipeline__logreg__C': 100.0, 'pipeline__sfs__k_features': 3, 'randomforestclassifier__n_estimators': 200}\n"
     ]
    }
   ],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, :], iris.target\n",
    "\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "from mlxtend.feature_selection import SequentialFeatureSelector\n",
    "\n",
    "clf1 = LogisticRegression(random_state=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "\n",
    "# Creating a feature-selection-classifier pipeline\n",
    "\n",
    "sfs1 = SequentialFeatureSelector(\n",
    "    clf1,\n",
    "    k_features=4,\n",
    "    forward=True,\n",
    "    floating=False,\n",
    "    scoring='accuracy',\n",
    "    verbose=0,\n",
    "    cv=0)\n",
    "\n",
    "clf1_pipe = Pipeline([('sfs', sfs1), ('logreg', clf1)])\n",
    "\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1_pipe, clf2, clf3], voting='soft')\n",
    "\n",
    "params = {\n",
    "    'pipeline__sfs__k_features': [1, 2, 3],\n",
    "    'pipeline__logreg__C': [1.0, 100.0],\n",
    "    'randomforestclassifier__n_estimators': [20, 200]\n",
    "}\n",
    "\n",
    "grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)\n",
    "grid.fit(iris.data, iris.target)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\" % (grid.cv_results_[cv_keys[0]][r],\n",
    "                                  grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "                                  grid.cv_results_[cv_keys[2]][r]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:55:04.072663Z",
     "start_time": "2018-04-19T09:55:04.066661Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'pipeline__logreg__C': 100.0,\n",
       " 'pipeline__sfs__k_features': 3,\n",
       " 'randomforestclassifier__n_estimators': 20}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grid.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:55:20.647123Z",
     "start_time": "2018-04-19T09:55:20.601469Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eclf = eclf.set_params(**grid.best_params_)\n",
    "eclf.fit(X, y).predict(X[[1, 51, 149]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:55:44.010718Z",
     "start_time": "2018-04-19T09:55:43.973916Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2])"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from mlxtend.feature_selection import ColumnSelector\n",
    "\n",
    "col_sel = ColumnSelector(cols=[0, 2])\n",
    "\n",
    "clf1_pipe = Pipeline([('sel', col_sel), ('logreg', clf1)])\n",
    "\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1_pipe, clf2, clf3], voting='soft')\n",
    "eclf.fit(X, y).predict(X[[1, 51, 149]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:56:00.719088Z",
     "start_time": "2018-04-19T09:56:00.686354Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done   4 out of   4 | elapsed:    0.0s finished\n",
      "Features: 1/2[Parallel(n_jobs=1)]: Done   3 out of   3 | elapsed:    0.0s finished\n",
      "Features: 2/2"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best features (0, 2)\n"
     ]
    }
   ],
   "source": [
    "sfs1 = SequentialFeatureSelector(\n",
    "    clf1,\n",
    "    k_features=2,\n",
    "    forward=True,\n",
    "    floating=False,\n",
    "    scoring='accuracy',\n",
    "    verbose=1,\n",
    "    cv=0)\n",
    "\n",
    "sfs1.fit(X, y)\n",
    "\n",
    "print('Best features', sfs1.k_feature_idx_)\n",
    "\n",
    "col_sel = ColumnSelector(cols=sfs1.k_feature_idx_)\n",
    "\n",
    "clf1_pipe = Pipeline([('sel', col_sel), ('logreg', clf1)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:56:22.127094Z",
     "start_time": "2018-04-19T09:56:22.092925Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2])"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eclf = EnsembleVoteClassifier(clfs=[clf1_pipe, clf2, clf3], voting='soft')\n",
    "eclf.fit(X, y).predict(X[[1, 51, 149]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 5 - Using Pre-fitted Classifiers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:57:05.111724Z",
     "start_time": "2018-04-19T09:57:05.080705Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, 1:3], iris.target\n",
    "\n",
    "\n",
    "from sklearn import model_selection\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import numpy as np\n",
    "\n",
    "clf1 = LogisticRegression(random_state=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "\n",
    "for clf in (clf1, clf2, clf3):\n",
    "    clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:57:22.078901Z",
     "start_time": "2018-04-19T09:57:22.063253Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy: 0.9733333333333334\n"
     ]
    }
   ],
   "source": [
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "import copy\n",
    "eclf = EnsembleVoteClassifier(clfs=[clf1, clf2, clf3], weights=[1, 1, 1], refit=False)\n",
    "\n",
    "labels = ['Logistic Regression', 'Random Forest', 'Naive Bayes', 'Ensemble']\n",
    "\n",
    "eclf.fit(X, y)\n",
    "\n",
    "print('accuracy:', np.mean(y == eclf.predict(X)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 6 - Ensembles of Classifiers that Operate on Different Feature Subsets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T09:58:06.960154Z",
     "start_time": "2018-04-19T09:58:06.938564Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "EnsembleVoteClassifier(clfs=[Pipeline(memory=None,\n",
       "     steps=[('columnselector', ColumnSelector(cols=(0, 2), drop_axis=False)), ('logisticregression', LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          penalty='l...='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False))])],\n",
       "            refit=True, verbose=0, voting='hard', weights=None)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.classifier import EnsembleVoteClassifier\n",
    "from mlxtend.feature_selection import ColumnSelector\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "iris = load_iris()\n",
    "X = iris.data\n",
    "y = iris.target\n",
    "\n",
    "pipe1 = make_pipeline(ColumnSelector(cols=(0, 2)), LogisticRegression())\n",
    "pipe2 = make_pipeline(ColumnSelector(cols=(1, 2, 3)), LogisticRegression())\n",
    "\n",
    "eclf = EnsembleVoteClassifier(clfs=[pipe1, pipe2])\n",
    "\n",
    "eclf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Logistic Regression\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/LogisticRegression_files/logistic_regression_schematic.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 1 - Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:00:59.904693Z",
     "start_time": "2018-04-19T10:00:59.263920Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 100/100 | Cost 0.32 | Elapsed: 0:00:00 | ETA: 0:00:00"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x113144a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1130d5438>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import LogisticRegression\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "lr = LogisticRegression(eta=0.1, \n",
    "                        l2_lambda=0.0, \n",
    "                        epochs=100,\n",
    "                        minibatches=1, # for Gradient Descent\n",
    "                        random_seed=1,\n",
    "                        print_progress=3)\n",
    "lr.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=lr)\n",
    "plt.title('Logistic Regression - Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(lr.cost_)), lr.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predicting Class Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:01:19.652824Z",
     "start_time": "2018-04-19T10:01:19.648090Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Last 3 Class Labels: [1 1 1]\n"
     ]
    }
   ],
   "source": [
    "y_pred = lr.predict(X)\n",
    "print('Last 3 Class Labels: %s' % y_pred[-3:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predicting Class Probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:01:37.540219Z",
     "start_time": "2018-04-19T10:01:37.534479Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Last 3 Class Labels: [0.99997968 0.99339873 0.99992707]\n"
     ]
    }
   ],
   "source": [
    "y_pred = lr.predict_proba(X)\n",
    "print('Last 3 Class Labels: %s' % y_pred[-3:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Stochastic Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:02:13.975252Z",
     "start_time": "2018-04-19T10:02:13.151650Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 30/30 | Cost 0.27 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112be0f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10a5b3198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import LogisticRegression\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "lr = LogisticRegression(eta=0.5, \n",
    "                        epochs=30, \n",
    "                        l2_lambda=0.0, \n",
    "                        minibatches=len(y), # for SGD learning \n",
    "                        random_seed=1,\n",
    "                        print_progress=3)\n",
    "lr.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=lr)\n",
    "plt.title('Logistic Regression - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(lr.cost_)), lr.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 3 - Stochastic Gradient Descent w. Minibatches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:02:51.740224Z",
     "start_time": "2018-04-19T10:02:51.233373Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 30/30 | Cost 0.25 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAEICAYAAABLdt/UAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XucHHWZ7/HPMzOd+/0GyeRGSAQCIioXgaOwKGtgI3hZXkcEFVHDAVw1ouJt9+DqKuhy8ay6JiuIeAGvYAAlQYHDcQUBEYUkBELIbcjFhCSTZEjS0/2cP341SWe6J90zXVM9XfN9v155Zbqq+ldPVVc9Xf2rp39t7o6IiKRHQ60DEBGReCmxi4ikjBK7iEjKKLGLiKSMEruISMoosYuIpIwSeydmdpGZLenhc5ea2Zkxh9TnmdlvzOz9tY6jt5jZJWb2+wTWM9XMdplZY2+vqyc674co1hm1jElKq+vEbmarzewtcbbp7j9y97+vYN23mtmXOz33WHd/qDvrM7PpZubRSbIr2qbPdDPsmnL3c9z9+0mv18wGmNn1Zra+YN/dVDA/9uMjTp3jc/e17j7M3XM9aGuimf2Xmb0U7YtV0TF6dLxRHxDFuqradkqdSyWWcTPbHW3bVjP7nZn9z2rX3VvM7Boz+2Gt1l/XiT1lRrn7MOAfgX82s7PjXoGZNcXdZo19FjgROBkYDpwJPFnLgGrBzMYCfwCGAG8k7IvXAf8XKHkc1emx8JroHDkKuBX4ppn979qG1Ee5e93+A1YDb+li3oeBlcDLwCJgUsG8vwdWADuAbxNOgA9F8y4Bfh/9bcCNwGagFXgaOA6YB2SBfcAu4O7O8QCNwOeAF4CdwJ+AKSXinA440FQw7THgUwWPJwG/AP4GvAh8tGDeYOD7wDZgOfBpYH2nfXQ18FdgL9BUpr2TgSei7d0E3BBNHwT8ENgKbAceBw6L5j1UsP8agC8Aa6L9dhswstO2vh9YC2wBPl/F638P8PEu5v0AyAOvRK/Rp6Pp5wFLo214CDim4DlTgF9G+2Ur8M3CYwL492g/vwicU/C8D0T7fiewCrisYN64KM7thGPx/0X7qCi+zscCMAb4HvBStN67utjWLwN/ARoOsa862v5gtO8fjqb/DNhIOBceBo4teM5YwrnTSjgmv0R0bkTzHZgZ/T0w2j9ro+PmO8DgaN6ZwHrgquiY2AB8IJpX8lwqEf/+dRVM+0dgDzA2ejwSuDlqvyXaL43RvJmE83xHdNz9pKCdY4H7o9dnE/C5gmP5M4RzeCvwU2BMuWMZmBNtTzbapr8knhuTXmGswXeR2IGzoh39uuiA+4+CA3lcdKC+k5DkPha9AKUS+1sJCXkUIckfA0yM5t0KfLmreIBPEd4Ijoqe+5qOA7CLE67jZH4D0Aa8o+Dg+hPwL8AAYAYhebw1mn9tdMCOBiYTEnjnxP4UIWkNrqC9R4D3Rn8PA94Q/X0ZcDfhqrAReD0wIpr3UMH+u5Twhjojev4vgR902tb/imJ5DeHN5phSr28Fr/8XopPqCuDVgB3q+ABeBewmXMVmCMl0ZbQfGgnJ8UZgKOGN7H8UHBNZwsVCI3A5IdlaNP8fgCOj1/mM6PV7XTTvq4Qkl4n+vbHgeZ3j63ws3Av8JHptM8AZXeyHR4FryuyrjrZvi7avI+leSrjCHwjcBDxV8Jw7CMlsKOGCpoWuE/uNhDeBMVF7dwNfjeadCbQD/xptx7nRPhrd1blUIv5SiT0TtXtO9PhOYEEU7wTCm9Fl0bzbgc8Tjv/C13Y44Y3gqmj6cOCUaN7Hon07Odo/C4DbKzmWgWuAH9YsN9ZqxbEE33Vivxn4WsHjYYQTczrwPuCRgnkGrKN0Yj8LeI6QbBs6raPoYOTgxL4COL+Cbeg4QLYTrt6ccOXTcfKfAqzt9JzPAt+L/t6flKPHH6I4sV9a8Lhcew8DXwTGdVrmUsLH/eNLbMNDBfvvd8AVBfOOivZ9U8G2Ti6Y/xjw7h6+/o3AlcB/RyfVS8D7uzo+gH8GflrwuIGQrM4ETiVcqTeVWM8lwMqCx0Oi7Ti8i7juAj4W/f2vwK/olJS6iK9j/zQBEwlX9KMr2A8rgf9V8Pi86HjaCSzp1PaMQ7QzKlpmZLRvs8DRBfO/QonETjiHdgNHFsw7FXgx+vtMwrFd+Kl0MwcuGm6lB4k9mr4RuAg4LDoGBhfMuxB4MPr7NmBh4bFXsMyfu1jncuDNBY8nVnosU+PEntY+9kmErgAA3H0X4aNUczRvXcE8J3xMLOLuDwDfBL4FbDazhWY2osIYphA+wlVqHOEN6CrCiZCJpk8DJpnZ9o5/hC6ew6L5B21Pp79LTSvX3gcJV7bPmtnjZjY3mv4DYDFwR3SD7mtmlqHYQfs++rupoH0IJ2OHtmi7D1JQIbLLzHaVWA/unnP3b7n76YSk9G/ALWZ2TKnlO8fm7nnCvmkmvF5r3L29i+duLHheW/TnsCjWc8zsUTN7Odqf5xJeT4CvExLvkuiGZqU3xqcAL7v7tgqW3UpIOh3xLXL3UcB8wqeRQvuPBTNrNLNrzewFM2slvNEQxT6e8LoVHjuFr2uh8YQ3uz8VHFP3RdP3x9hp35Z83bsjOv7GE7pQphHOmQ0FMSwgXLlD+HRmwGNR9dql0fRDnafTgDsL2lsO5OjmsVwLaU3sLxFeFADMbCihv7CF8LFrcsE8K3zcmbv/H3d/PTCbkPA+1TGrTAzrCB/PKxYlqhsI/YZXFLTzoruPKvg33N3PjeYftD2EA7Wo6U5xddmeuz/v7hcSTojrgJ+b2VB3z7r7F919NnAaMJfw6aezg/Y9MJXwcXlTN3YFfqBCZJiHG2blln/F3b9F6IueXWK7i2KLXvsphONiHTC1uzcVzWwg4X7FvxPuOYwCfk1IIrj7Tne/yt1nEK6kP2Fmb+4ivkLrgDFmNqqCMH4HvN3MKjmfC9f5HuB84C2Eq/TpHZtF+PTSzsHH09Qu2txCuCI/tuCYGlnJ61Yipu44P4rxMcL+2kv4pNkRwwh3PxbA3Te6+4fdfRKhW/HbZjYzel5XJZvrCN08hefKIHdv6cVtikUaEnvGzAYV/Gsi9Kd9wMxOiE68rwB/dPfVhH7LV5vZ26NlrwQOL9WwmZ1kZqdEVwa7CQk3H83eRNcHBMB3gS+Z2SwLjo+qFypxLfBpMxtEOGh3mtnVZjY4uso6zsxOipb9KfBZMxttZs3AR8q0fcj2zOxiMxsfXc1uj56TN7O/M7NXW6ixbiV8JM2XaP92YL6ZHWFmwwj7/ieHuBLuMTP7uJmdGW1Hk4Va+uHAn6NFOr9GPwX+wczeHL2mVxGSwR8I+2UDcK2ZDY2OpdMrCGMAof/1b0C7mZ1DuDnfEeNcM5sZvYnsIFzxlT2G3H0D8BtCAhptZhkze1MXMdxA6If/gZkdGR1vw4ETysQ+nLD9WwlX3F8pWH+OcH/kGjMbYmazCTcKS8WaJ/Q132hmE6Ltbjazt5ZZf4dy59JBzGyMmV1E+CR9nbtvjfbXEuB6MxthZg3Rvjgjes4FZtZxAbSNkHjzhBvbE6NjaaCZDTezU6LlvgP8m5lNi9oYb2bnd2Obplf4Zhu7NCT2XxOuFjr+XePuvyX0p/6CcLIeCbwbwN23ABcAXyMc0LMJVSB7S7Q9gnDAbiN8DN1K+GgNoR9/dvQx7a4Sz72BkEiWEBLhzYSbLJW4N1rnh6MTbC7hJH2RcHX0XcIVFoQ+3PXRvN8CP+9iW4D9J+yh2psDLI26P75B6DN8hfDm9/NoW5YTbtj+oMQqbommPxy1vwf4pwq3u7vagOsJH4e3EN6k3+UHaqu/Cnwheo0+6e4rgIsJN9O3AG8D3ubu+6L98jZCn/Fawj4tWyft7juBjxJe622Eq+BFBYvMIrwuuwg3pr/t7g+Wiq9E8+8lvIE+S+iT/ngXMWwh3AfaQ6je2Um4YT6ccKO3K7cRjusWYBnhRmGhjxC6FjYS+sG/d4i2riZ0OT0adev8lnB/pRLlzqUOf4mOy5WEe0nz3f1fCua/j/BGu4zwWvycA11UJwF/jJ6/iHAPZFX0+p1NeO03As8Dfxc95xvRskvMbCdh/3Qk/XJ+Fv2/1cwSL8HtuEHXb0XvqOuBiwpOuLplZpcTkvEZtY5FRGojDVfs3WZmbzWzUVE3zecIfYqdr1bqgoVvHJ4effQ8itC9cGet4xKR2qnHb5/F4VTgxxz42Pb2qLuhHg0g3P0/gtAnfgfhS1ci0k/1+64YEZG06ZddMSIiaVaTrpgnn6xtjaeISL0ZPhxmzQrfjyinJom9pZLyfhER2W/sWJg1q7Jl1RUjIpIySuwiIimjxC4ikjL9tY5dRPqlPEOG7CaT6favDyYqm22krW0oPb32VmIXkX5jyJDdDBuWIYxPV1GBSQ04AwfuBXbT1ja8Ry2oK0ZE+o1MJkcYSaSvJnUAw2xgVZ8qlNhFpJ/py0m9Q3UxKrGLiKSMEruISMIeeeQ+LrjgKN71rpl8//vXxt6+EruISIJyuRxf//qV3HTTb7jjjmUsWXI7q1Yti3UdqooRESnhny45ibZtm4umDxk9gf+49fEet7ts2WNMnjyT5ubwa4Bnn/1uHn74V8yYMbvMMyunxC4iUkLbts08Omp80fQ3lEj23bF5cwuHHXbgN8InTJjM0qV/rKrNztQVIyKSMkrsIiIJmjChmU2b1u1/vHnzesaPb451HUrsIiIJOuaYk1i37nleeulFstl93H//HbzpTefFug71sYuIJKipqYlPfvKbfPSjbyWfz/G2t13KjBnHxruOWFsTEUmJIaMnlLxROmT0hKrbPv30czn99HOrbqcrSuwiIiVUU9JYa+pjFxFJGSV2EZGUUWIXEUkZJXYRkZRRYhcRSRkldhGRhH3pS5cyZ84ELrzwuF5pX4ldRCRhc+dewk033ddr7Suxi4gcwvbtW5g//13s2LE1tjZf+9o3MWLEmNja60yJXUTkEO666zaWLt3GnXd+v9ahVEyJXUSkC9u3b+Hee+9hzJj/5N5774n1qr03KbGLiHThrrtuI5eby8CBR5HLza2bq3YldhGREjqu1gcPfh8Agwe/r26u2qtO7GY2xcweNLNlZrbUzD4WR2AiIrXUcbXe1DQOgKamcbFdtX/hCxfyoQ+dypo1K5g7dzKLFt1cdZuF4hjdsR24yt2fNLPhwJ/M7H53j/dnt0VEEvTYYw+Ry73E9u0/7jR9Epdc8omq2v7yl2+v6vnlVJ3Y3X0DsCH6e6eZLQeaASV2Ealb3/72olqH0GOx9rGb2XTgtUDRT26b2Twze8LMnrjvvoVxrlZERArE9kMbZjYM+AXwcXdv7Tzf3RcCCwHuvhuPa70iIt3jgNU6iDKqS5GxXLGbWYaQ1H/k7r+Mo00Rkbhls42476XaxNm7HPe9ZLONPW6h6it2MzPgZmC5u99QbXsiIr2lrW0osJtMZk+tQzmkbLYxirVn4uiKOR14L/C0mT0VTfucu/86hrZFRGLUQFvb8FoH0eviqIr5PX2/w0pEpN/QN09FelFr6xa++MV30dra97+tKOmhxC7SixYvvo3nntvG4sX1McaIpIMSu0gvaW3dwgMP3MPo0f/JAw/co6t2SYwSu0gvWbw4jDUyYEAYGVBX7ZIUJXaRXtBxtV44MqCu2iUpSuwivaDjar2xMYwM2Ng4TlftkpjYhhQQkQOeeiqMDLhjx487TZ/EBRdUNzKgSDnmnvxXazVWjEj/1Nq6hRtvvIz58xcyYsTYWodTV8aOhdNOq+w7Q+qKEZHEqPwzGUrsIpIIlX8mR4ldRBKh8s/kKLGLSK9T+WeylNhFpNep/DNZKncUkV6n8s9kqdxRRKQOqNxRpI60tKzg4ouPoKXl+V5bh4YP7l+U2EVqbMGCz9DaOoYFCz7da+tQ/Xj/osQuUkMtLSt45pmnyWS+xzPPPN0rV+2qH+9/lNhFamjBgs+Qz7+Hhobjyeff0ytX7aof73+U2EVqpONqvbHxcgAaGy+P/apd9eP9kxK7SI0cuFqfCEBDw8TYr9pVP94/qY5dpEaef/7PwGNkszd3mp6JbR2qH++fVMcuIlIHVMcuqRdHXXYStd2VrCOJbal2flJxJtVG2imxS12Koy47idruStaRxLZUOz+pOJNqI+2U2KXuxFGXnURtdyXrSGJbqp2fVJxJtdEfKLFL3YmjLjuJ2u5K1pHEtlQ7P6k4k2qjP1Bil7oSR112ErXdlawjiW2pdn5ScSbVRn+hxC51JY667CRquytZRxLbUu38pOJMqo3+QnXsUlfiqMtOora7knUksS3Vzk8qzqTa6C9Uxy6Ja23dwo03Xsb8+QsZMWJsrcPp81paVnD11XO47rolNDfPqnU4XarkdS23TBzHRlqPL9WxS5+mcrXuSWJY3zjEUTKpcsh4KLFLolSu1j1JDOsbhzhKJlUOGR8ldkmUytW6J4lhfeMQR8mkyiHjE0tiN7NbzGyzmT0TR3uSTipX654khvWNQxwlkyqHjFdcV+y3AnNiaktSSuVq3ZPEsL5xiKNkUuWQ8Yql3NHdHzaz6XG0JemlcrXuSWJY3zjEUTKpcsh4xVbuGCX2e9z9uC7mzwPmAVxxxYLXz5kzL5b1ioj0B32y3NHdF7r7ie5+opK6JKGlZQUXX3xEl33SK1Y8yjveMYYVKx7rso1KlultcQypG8ewvXHEKclQVYykVrn67+uvv5xcbjrXX39Zl21UskxvS6I+XPXj6aLELqlUrv57xYpH2bhxG3ALGzduK3lFXskyvS2J+nDVj6dPXOWOtwOPAEeZ2Xoz+2Ac7Yr0VLn67+uvvxy4CLMTgItKXpFXskxvS6I+XPXj6RNLYnf3C919ortn3H2yu99c/lkivaNc/feBK/Ero2dcWXRFXskyvS2J+nDVj6eTumIkdcrVfx+4Ep8EEP1/8BV5Jcv0tiTqw1U/nk4atldSp1z99+bN64Cb6fzBcvPm9oK/yy/T25KoD1f9eDpp2F4RkTrQJ+vYRZKWRG12uVr5OKg+XLpLiV1SK4na7CTGSld9uHSXErukUhK12UmMla76cOkJJXZJpSRqs5MYK1314dITSuySOknUZicxVrrqw6WnlNgldZKozU5irHTVh0tPqY5dUieJ2uwkxkpXfbj0lOrYRUTqgOrYpab6St11EmOQi/RFSuwSu75Sd53EGOQifZESu8Sqr9RdJzEGuUhfpcQuseordddJjEEu0lcpsUts+krddRJjkIv0ZUrsEpu+UnedxBjkIn2Z6tglNn2l7jqJMchF+jLVsctBWlu3cOONlzF//kJGjBhbcpmWlhVcffUcrrtuCc3NsxKOsHKVbEu1bSSxDhFQHbtUoZISwCSGqo1DEsP2JrEOke5SYpf9KikBTGKo2jgkMWxvEusQ6QkldtmvkhLAJIaqjUMSw/YmsQ6RnlBiF6CyEsAkhqqNQxLD9iaxDpGeUmIXoLISwCSGqo1DEsP2JrEOkZ6qSbnjiZm/8AqDWZadhVd2k1d6WSUlgEkMVRuHJIbtTWIdIj1Vk3JH7r7bAbI2oCix56yJvzacQFt+0EHTX8kPJEdjcjGKiPQh3Sl3rOkXlDK+r3ii7+WU/H8XTd7WNI51Nq1o+k4fxsvtI3ojPBGRulQ33zwd3b6F0Wwpmp6ngY2ZKUVX/nsZyPLsTPK6jSAi/UzdJPauNJBnUnZNyXmTbTV5O7j7Jk8DzzQcz8780KLlX8kPVJ+/iNS9uk/shzLA91Jq8IKT838oufyGzFQ2cVjR9G35kezKDY47PBGRXpHqxN5dE7NrmcjaoulZG8DGzGQK7zO3W4Zns0fqhq6I9DlK7BXI+D6mZFcVTW+2NbRbcanfssbj2J4rvqG7N59RV4+I9Dol9ioM9D0M9D1F00/KP1Jy+Y1Nk2mhuWj6Th/GztyQ2OMTkf5JiT1Bh7ev53DWF01vtwzrM9OLpu9hEM9ni6eLiByKEnsf0ORZpmdLj7cyteFFcnbwy5Sjkb9yfNGXuNq9UV09IhJPYjezOcA3gEbgu+5+bRztCgzJ7yo5/U08WDTtb00TWW3Ti6bvzg9RV49IP1J1YjezRuBbwNnAeuBxM1vk7suqbVu6Z3z7BsazoWh6jkbWZo4suprfxwBWZqfpKl8kZeK4Yj8ZWOnuqwDM7A7gfECJvY9oJMcR2edKzpvSsJosB1f2eEMDz/hxJa/yVd4p0vfFkdibgXUFj9cDp3ReyMzmAfMAFlxxBfPmzIlh1VKtofmdxRPz8EYeKrn8msxMNuSLv8S104exJz8g5uhEpCcSu3nq7guBhcD+0R2l/kzLrmQaK4um72kYwrpOlT05GnkhO1Xj9YgkLI7E3gJMKXg8OZom/cigfBuz8sW9b5Mb17LPBhZNX84xvJwdXtyQqb9fpFpxJPbHgVlmdgQhob8beE8M7UoKDMvtKDn9dDZT6p7t+swRrM1PLpre5oN5JV/8BiEixapO7O7ebmYfARYTyh1vcfelVUcm/dLk7ItM5sWi6ftsIKszM4sqePYwiLXZiUmFJ1IXYuljd/dfA7+Ooy2RUgb4Xl6VLX29MK1xbNGYPXkaeDp/bNGXuET6A33zVOreqNzWktPPYgOdf/lx44CprPLpRcvuyQ9UV4+khhK7pJbhRfdiJ2XXMIniH2bJ2gBeyBxV8ktc6uqReqPELkIYmvno7NMl501tGsc+P7hG362B5X60foBF+iQldpEyRrcX/9YuwAReKvrpRYAXm2bxUq74S1yv+CCyrlNOep+OMpEeaiBPg+eLps/KLmNWiRE1djcMZ3XTkQf1++esiXXZwzVej8RKiV0qdtYnPkHrjuK69BEjR/LADTfUIKL6MjS/k2PzTxVNn9o0jj1WPC7Pc/mZtOaKf3RdpBwldqlY644dPDFyZNH0E0ske6lcV109h7G+5E8vrm+azpr8lKLpe/MZdfUIoMQu0mc1kGeA7y2aPiO7ghmsKJr+SsNQnm88qmhYhr0MYGN2XK/FKX2PErtISgzO7+b4/JMl521qai75Ja7l7bPY6xqVM22U2EX6gcPaS4/LN9HWke2U8Dc1NfNCbnrRsllvUldPndCrJNKPNXmWJrIHTZuefZ7pFP8G714bxLOZ48j7wV097dakrp4+RoldKjZi5MiSN0pHlLihKukz0PfwmuwTJedtampmrxWPy/Nc7kgN1VAD5p0H00iCfmhDpF/I2oCS4/GvbjySlvbiL3FlvUk/zNKFsWPhtNMq+8KDrtglNqpzl84yvo+M7yuafmz+KY4tsXxr42hWNrzqoC9x5a1BXT3dpMQusVGdu1RrRG4br8v9sWj6pqZmXinxJa5VuWnszmu8ns6U2EWkz+uqqqfZ1rCnsTjhtzRMYXV78S9x5byhX3T1KLGLSN3K+D4yueKunqNz2zma4tE6dzWO5NmG2XS+ybfPB/By+4heijJ5Suwi0m8My+3gxNwjJeetzRxJe6eUmKeBle3T665+v76iFRHpJVOzL5ScPrlhDXs6lXJubpzIyuy0omXzNPSJkTqV2CU2qnOXNBqUb2MQbQdNG5V/mVdR/Bu8bQ3DWNr4anL5g/vx262JbQl29SixS8XKlTOuWLcOstmi+RtaW2OLYez555Mp8d2LrBlbf/Wr2Naj0k3piSH5XZyUL93Vsy4zgz0Uf4lrVftU9nnxKJ7VUGKXipUtZ8xmackUH6DNJZJ9T2Xc2dj5h0yBw2P+op1KNyVuU7KrSk9vWE1b47Ci6WtsOuv3Tdj/2Eoc911RYhcRqaFB+TYG5duKpo9hM68tzOU+FjitojbTX9ApItLPKLGLiKSMEruISMqoj10qVracMZMpfaO0xA3VnsqalbxRmu3GjaVKqHRT6pmG7RURqQdh3F4N21svytVMx1FTnUQbzRdcULKOnUyGlp/9rKJ19BWqY5d6psTeB5SrmY6jpjqRNhKoY0+K6tilnunmqYhIyiixi4ikjBK7iEjKKLGLiKRMVTdPzewC4BrgGOBkd38ijqD6m3I103HUVCfSRgJ17ElRHbvUs6rq2M3sGCAPLAA+WXFiVx27iEj3JFXH7u7LoXvDSfY3cYwfXq6NSurHy7VRSd326PPOY2CJ+PYC2xYtiiXOOGrlk6inV5279GWJ1bGb2TxgHsCCK65g3pw5Sa26puIYP7xsGxXUj5dro5K67YHAxhLxHR5jnLHUyidQT686d+nLyiZ2M/stB87dQp9394p/ssbdFwILAXXFiIj0orKJ3d3fkkQgIiISD5U7ioikTFWJ3czeYWbrgVOBe81scTxhiYhIT1VbFXMncGdMsaRSHOOHl22jgvrxcm1UUre9l9I3W/bGGGcstfIJ1NOrzl36Mo3HXqV6KXsrF2ccZZnVxiAih6Dx2JNTL2Vv5eKMoyyz2hhEJB66eSoikjJK7CIiKaPELiKSMkrsIiIpo5unVaqXsrdyccZRllltDCISD5U7iojUg26UO6orRkQkZZTYRURSRoldRCRllNhFRFJGiV1EJGWU2EVEUkaJXUQkZZTYRURSRoldRCRllNhFRFJGiV1EJGWU2EVEUkaJXUQkZZTYRURSRoldRCRllNhFRFJGiV1EJGWU2EVEUkaJXUQkZZTYRURSRoldRCRllNhFRFJGiV1EJGWU2EVEUkaJXUQkZZTYRURSpqrEbmZfN7NnzeyvZnanmY2KKzAREemZaq/Y7weOc/fjgeeAz1YfkoiIVKOqxO7uS9y9PXr4KDC5+pBERKQacfaxXwr8pquZZjbPzJ4wsycW3ndfjKsVEZFC5u6HXsDst8DhJWZ93t1/FS3zeeBE4J1erkGAu+8uv4yIiBwwdiycdppVsmhTuQXc/S2Hmm9mlwBzgTdXlNRFRKRXlU3sh2Jmc4BPA2e4e1s8IYmISDWq7WP/JjAcuN/MnjKz78QQk4iIVKGqK3Z3nxlXICIiEg9981REJGWU2EVEUkaJXUQkZarqY++xkSNrsloRkbo1bFjFi5b9gpIEZjbP3RfWOo64pXG70rhNoO2qN7XcLnXFVG5erQPoJWncrjRuE2i76k3NtkuJXUQkZZTYRURSRom9cqnrA4ykcbvSuE2g7ao3Ndsu3TwVEUnqZuZkAAACN0lEQVQZXbGLiKSMEruISMoosXdDGn+828wuMLOlZpY3sxNrHU+1zGyOma0ws5Vm9plaxxMHM7vFzDab2TO1jiUuZjbFzB40s2XR8fexWscUBzMbZGaPmdlfou36Yi3iUGLvnjT+ePczwDuBh2sdSLXMrBH4FnAOMBu40Mxm1zaqWNwKzKl1EDFrB65y99nAG4ArU/Ja7QXOcvfXACcAc8zsDUkHocTeDWn88W53X+7uK2odR0xOBla6+yp33wfcAZxf45iq5u4PAy/XOo44ufsGd38y+nsnsBxorm1U1fNgV/QwE/1LvEJFib3nDvnj3VITzcC6gsfrSUGySDszmw68FvhjbSOJh5k1mtlTwGbgfndPfLtqMwhYH9aNH+9uB36UZGw9Vck2idSCmQ0DfgF83N1bax1PHNw9B5wQ3YO708yOc/dE748osXeSxh/vLrdNKdICTCl4PDmaJn2QmWUISf1H7v7LWscTN3ffbmYPEu6PJJrY1RXTDQU/3n2efry7T3ocmGVmR5jZAODdwKIaxyQlmJkBNwPL3f2GWscTFzMb31EtZ2aDgbOBZ5OOQ4m9e1L3491m9g4zWw+cCtxrZotrHVNPRTe2PwIsJtyM+6m7L61tVNUzs9uBR4CjzGy9mX2w1jHF4HTgvcBZ0bn0lJmdW+ugYjAReNDM/kq40Ljf3e9JOggNKSAikjK6YhcRSRkldhGRlFFiFxFJGSV2EZGUUWIXEUkZJXYRkZRRYhcRSZn/D8eFceVn/gk5AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112be0940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x110cf25c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import LogisticRegression\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "lr = LogisticRegression(eta=0.5, \n",
    "                        epochs=30, \n",
    "                        l2_lambda=0.0, \n",
    "                        minibatches=5, # 100/5 = 20 -> minibatch-s \n",
    "                        random_seed=1,\n",
    "                        print_progress=3)\n",
    "lr.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=lr)\n",
    "plt.title('Logistic Regression - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(lr.cost_)), lr.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural Network - Multilayer Perceptron\n",
    "**Basic Architecture**\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/NeuralNetMLP_files/neuralnet_mlp_1.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 1 - Classifying Iris Flowers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:06:17.605190Z",
     "start_time": "2018-04-19T10:06:17.596959Z"
    }
   },
   "outputs": [],
   "source": [
    "from mlxtend.data import iris_data\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]]    \n",
    "\n",
    "# standardize training data\n",
    "X_std = (X - X.mean(axis=0)) / X.std(axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:06:37.295705Z",
     "start_time": "2018-04-19T10:06:36.917337Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 150/150 | Cost 0.06 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    }
   ],
   "source": [
    "from mlxtend.classifier import MultiLayerPerceptron as MLP\n",
    "\n",
    "nn1 = MLP(hidden_layers=[50], \n",
    "          l2=0.00, \n",
    "          l1=0.0, \n",
    "          epochs=150, \n",
    "          eta=0.05, \n",
    "          momentum=0.1,\n",
    "          decrease_const=0.0,\n",
    "          minibatches=1, \n",
    "          random_seed=1,\n",
    "          print_progress=3)\n",
    "\n",
    "nn1 = nn1.fit(X_std, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:08:45.598641Z",
     "start_time": "2018-04-19T10:08:45.180783Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1135dd160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.plotting import plot_decision_regions\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "fig = plot_decision_regions(X=X_std, y=y, clf=nn1, legend=2)\n",
    "plt.title('Multi-layer Perceptron w. 1 hidden layer (logistic sigmoid)')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:09:05.242427Z",
     "start_time": "2018-04-19T10:09:05.049426Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112acdf98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(range(len(nn1.cost_)), nn1.cost_)\n",
    "plt.ylabel('Cost')\n",
    "plt.xlabel('Epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:09:19.218772Z",
     "start_time": "2018-04-19T10:09:19.214086Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 96.67%\n"
     ]
    }
   ],
   "source": [
    "print('Accuracy: %.2f%%' % (100 * nn1.score(X_std, y)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Stochastic Gradient Descent\n",
    "\n",
    "Setting minibatches to n_samples will result in stochastic gradient descent training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:10:13.357233Z",
     "start_time": "2018-04-19T10:10:13.048039Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 5/5 | Cost 0.11 | Elapsed: 00:00:00 | ETA: 00:00:00"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10987b048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "nn2 = MLP(hidden_layers=[50], \n",
    "          l2=0.00, \n",
    "          l1=0.0, \n",
    "          epochs=5, \n",
    "          eta=0.005, \n",
    "          momentum=0.1,\n",
    "          decrease_const=0.0,\n",
    "          minibatches=len(y), \n",
    "          random_seed=1,\n",
    "          print_progress=3)\n",
    "\n",
    "nn2.fit(X_std, y)\n",
    "\n",
    "plt.plot(range(len(nn2.cost_)), nn2.cost_)\n",
    "plt.ylabel('Cost')\n",
    "plt.xlabel('Epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:10:45.610023Z",
     "start_time": "2018-04-19T10:10:45.191756Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 25/25 | Cost 0.07 | Elapsed: 0:00:00 | ETA: 0:00:000"
     ]
    }
   ],
   "source": [
    "nn2.epochs = 25\n",
    "nn2 = nn2.fit(X_std, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:11:22.152168Z",
     "start_time": "2018-04-19T10:11:21.965425Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10987b358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(range(len(nn2.cost_)), nn2.cost_)\n",
    "plt.ylabel('Cost')\n",
    "plt.xlabel('Epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Classifying Handwritten Digits from a 10% MNIST Subset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:14:43.432650Z",
     "start_time": "2018-04-19T10:14:40.254233Z"
    }
   },
   "outputs": [],
   "source": [
    "from mlxtend.data import mnist_data\n",
    "from mlxtend.preprocessing import shuffle_arrays_unison\n",
    "\n",
    "X, y = mnist_data()\n",
    "X, y = shuffle_arrays_unison((X, y), random_seed=1)\n",
    "X_train, y_train = X[:500], y[:500]\n",
    "X_test, y_test = X[500:], y[500:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualize a sample from the MNIST dataset to check if it was loaded correctly:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:14:48.465389Z",
     "start_time": "2018-04-19T10:14:48.303303Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1130badd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_digit(X, y, idx):\n",
    "    img = X[idx].reshape(28,28)\n",
    "    plt.imshow(img, cmap='Greys',  interpolation='nearest')\n",
    "    plt.title('true label: %d' % y[idx])\n",
    "    plt.show()\n",
    "\n",
    "plot_digit(X, y, 3500)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Standardize pixel values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:14:52.134918Z",
     "start_time": "2018-04-19T10:14:51.938520Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from mlxtend.preprocessing import standardize\n",
    "\n",
    "X_train_std, params = standardize(X_train, columns=range(X_train.shape[1]), return_params=True)\n",
    "X_test_std = standardize(X_test, columns=range(X_test.shape[1]), params=params)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Initialize the neural network to recognize the 10 different digits (0-10) using 300 epochs and mini-batch learning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:15:47.339640Z",
     "start_time": "2018-04-19T10:15:47.332084Z"
    }
   },
   "outputs": [],
   "source": [
    "nn1 = MLP(hidden_layers=[150], \n",
    "          l2=0.00, \n",
    "          l1=0.0, \n",
    "          epochs=100, \n",
    "          eta=0.005, \n",
    "          momentum=0.0,\n",
    "          decrease_const=0.0,\n",
    "          minibatches=100, \n",
    "          random_seed=1,\n",
    "          print_progress=3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Learn the features while printing the progress to get an idea about how long it may take."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:16:21.357182Z",
     "start_time": "2018-04-19T10:16:07.482045Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 100/100 | Cost 0.01 | Elapsed: 0:00:13 | ETA: 0:00:00"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10ab9cc50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "nn1.fit(X_train_std, y_train)\n",
    "\n",
    "plt.plot(range(len(nn1.cost_)), nn1.cost_)\n",
    "plt.ylabel('Cost')\n",
    "plt.xlabel('Epochs')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:16:30.909437Z",
     "start_time": "2018-04-19T10:16:30.849306Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Accuracy: 100.00%\n",
      "Test Accuracy: 84.62%\n"
     ]
    }
   ],
   "source": [
    "print('Train Accuracy: %.2f%%' % (100 * nn1.score(X_train_std, y_train)))\n",
    "print('Test Accuracy: %.2f%%' % (100 * nn1.score(X_test_std, y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:18:07.666863Z",
     "start_time": "2018-04-19T10:18:07.663471Z"
    }
   },
   "source": [
    "# Perceptron\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/Perceptron_files/perceptron_schematic.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:18:47.904711Z",
     "start_time": "2018-04-19T10:18:47.901714Z"
    }
   },
   "source": [
    "## Example 1 - Classification of Iris Flowers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:19:02.853597Z",
     "start_time": "2018-04-19T10:19:02.262392Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 5/5 | Elapsed: 00:00:00 | ETA: 00:00:00TA: 00:00:00"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1130925f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Bias & Weights: [[-0.04500809]\n",
      " [ 0.11048855]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112c95e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import Perceptron\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "X = X[0:100] # class 0 and class 1\n",
    "y = y[0:100] # class 0 and class 1\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "\n",
    "# Rosenblatt Perceptron\n",
    "\n",
    "ppn = Perceptron(epochs=5, \n",
    "                 eta=0.05, \n",
    "                 random_seed=0,\n",
    "                 print_progress=3)\n",
    "ppn.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=ppn)\n",
    "plt.title('Perceptron - Rosenblatt Perceptron Rule')\n",
    "plt.show()\n",
    "\n",
    "print('Bias & Weights: %s' % ppn.w_)\n",
    "\n",
    "plt.plot(range(len(ppn.cost_)), ppn.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Missclassifications')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:19:54.709265Z",
     "start_time": "2018-04-19T10:19:54.706106Z"
    }
   },
   "source": [
    "# Softmax Regression\n",
    "**Logistic Regression model**\n",
    "![Logistic Regression model](http://rasbt.github.io/mlxtend/user_guide/classifier/SoftmaxRegression_files/logistic_regression_schematic.png)\n",
    "\n",
    "**Softmax Regression**\n",
    "![Softmax Regression](http://rasbt.github.io/mlxtend/user_guide/classifier/SoftmaxRegression_files/softmax_schematic_1.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:21:36.707645Z",
     "start_time": "2018-04-19T10:21:36.703218Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "y = np.array([0, 1, 2, 2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:21:43.232642Z",
     "start_time": "2018-04-19T10:21:43.226545Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "one-hot encoding:\n",
      " [[1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]]\n"
     ]
    }
   ],
   "source": [
    "y_enc = (np.arange(np.max(y) + 1) == y[:, None]).astype(float)\n",
    "\n",
    "print('one-hot encoding:\\n', y_enc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:21:50.104808Z",
     "start_time": "2018-04-19T10:21:50.087251Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Inputs X:\n",
      " [[ 0.1  0.5]\n",
      " [ 1.1  2.3]\n",
      " [-1.1 -2.3]\n",
      " [-1.5 -2.5]]\n",
      "\n",
      "Weights W:\n",
      " [[0.1 0.2 0.3]\n",
      " [0.1 0.2 0.3]]\n",
      "\n",
      "bias:\n",
      " [0.01 0.1  0.1 ]\n"
     ]
    }
   ],
   "source": [
    "X = np.array([[0.1, 0.5],\n",
    "              [1.1, 2.3],\n",
    "              [-1.1, -2.3],\n",
    "              [-1.5, -2.5]])\n",
    "\n",
    "W = np.array([[0.1, 0.2, 0.3],\n",
    "              [0.1, 0.2, 0.3]])\n",
    "\n",
    "bias = np.array([0.01, 0.1, 0.1])\n",
    "\n",
    "print('Inputs X:\\n', X)\n",
    "print('\\nWeights W:\\n', W)\n",
    "print('\\nbias:\\n', bias)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:22:19.177725Z",
     "start_time": "2018-04-19T10:22:19.171848Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "net input:\n",
      " [[ 0.07  0.22  0.28]\n",
      " [ 0.35  0.78  1.12]\n",
      " [-0.33 -0.58 -0.92]\n",
      " [-0.39 -0.7  -1.1 ]]\n"
     ]
    }
   ],
   "source": [
    "def net_input(X, W, b):\n",
    "    return (X.dot(W) + b)\n",
    "\n",
    "net_in = net_input(X, W, bias)\n",
    "print('net input:\\n', net_in)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:22:29.524643Z",
     "start_time": "2018-04-19T10:22:29.517910Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "softmax:\n",
      " [[0.29450637 0.34216758 0.36332605]\n",
      " [0.21290077 0.32728332 0.45981591]\n",
      " [0.42860913 0.33380113 0.23758974]\n",
      " [0.44941979 0.32962558 0.22095463]]\n"
     ]
    }
   ],
   "source": [
    "def softmax(z):\n",
    "    return (np.exp(z.T) / np.sum(np.exp(z), axis=1)).T\n",
    "\n",
    "smax = softmax(net_in)\n",
    "print('softmax:\\n', smax)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:22:39.175687Z",
     "start_time": "2018-04-19T10:22:39.170310Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "predicted class labels:  [2 2 0 0]\n"
     ]
    }
   ],
   "source": [
    "def to_classlabel(z):\n",
    "    return z.argmax(axis=1)\n",
    "\n",
    "print('predicted class labels: ', to_classlabel(smax))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:22:47.693935Z",
     "start_time": "2018-04-19T10:22:47.686013Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cross Entropy: [1.22245465 1.11692907 1.43720989 1.50979788]\n"
     ]
    }
   ],
   "source": [
    "def cross_entropy(output, y_target):\n",
    "    return - np.sum(np.log(output) * (y_target), axis=1)\n",
    "\n",
    "xent = cross_entropy(smax, y_enc)\n",
    "print('Cross Entropy:', xent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:22:57.561158Z",
     "start_time": "2018-04-19T10:22:57.553773Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost:  1.3215978715930938\n"
     ]
    }
   ],
   "source": [
    "def cost(output, y_target):\n",
    "    return np.mean(cross_entropy(output, y_target))\n",
    "\n",
    "J_cost = cost(smax, y_enc)\n",
    "print('Cost: ', J_cost)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 1 - Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:23:23.056122Z",
     "start_time": "2018-04-19T10:23:21.657035Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 500/500 | Cost 0.06 | Elapsed: 0:00:00 | ETA: 0:00:00"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x122b312e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xl0XeV57/Hvc0bNsmzJgGeDDcQYbEA2SaA0IQmBhELSkmLS9NImt3QIt23oBM1dpCWrvRnapGnCbaArrDZdJDRDE7xyaQgECJnAlsEYbGMjDMaWJ3mWNZ9znvvH3pKPj490ZFtbR8Pvs9ZZZ+93D3q2Ef753cO7zd0REREZTqzcBYiIyPinsBARkZIUFiIiUpLCQkRESlJYiIhISQoLEREpSWEhIiIlKSxERKQkhYWIiJSUKHcBo6WxsdEXLFhQ7jJERCaUdevW7Xf3plLrTZqwWLBgAS0tLeUuQ0RkQjGz7SNZT6ehRESkJIWFiIiUpLAQEZGSFBYiIlKSwkJEREpSWIiISEkKCxERKWnKh0Vnb4YvPL6VF948VO5SRETGrSkfFr2ZHP/841fZsPNIuUsRERm3pnxYJOMGQH82V+ZKRETGL4VFPPgj6FNYiIgMSWERhkV/xstciYjI+DXlwyIeM2IGmZx6FiIiQ5nyYQFB70KnoUREhqawIAgLnYYSERmawoLgjijdDSUiMjSFBUHPQtcsRESGFmlYmNl1ZrbFzFrN7K4iy//AzF4ys/Vm9jMzW5K37O5wuy1m9t4o60zGY/TpNJSIyJAiCwsziwP3AdcDS4Bb88Mg9A13v9jdlwOfA74QbrsEWAVcBFwH/N9wf5HQaSgRkeFF2bNYCbS6+zZ37wMeBm7KX8Hdj+bNVgMD/7y/CXjY3Xvd/XWgNdxfJHQaSkRkeIkI9z0b2JE3vxO4onAlM/s4cCeQAq7J2/bZgm1nR1OmTkOJiJRS9gvc7n6fu58H/BXwv09lWzO73cxazKylvb39tGvQaSgRkeFFGRZtwNy8+Tlh21AeBj5wKtu6+wPu3uzuzU1NTaddaDIeU1iIiAwjyrBYCyw2s4VmliK4YL06fwUzW5w3+37g1XB6NbDKzNJmthBYDKyJqtBkPEYmq9NQIiJDieyahbtnzOwO4DEgDjzo7hvN7F6gxd1XA3eY2buBfuAQcFu47UYz+xawCcgAH3f3bFS1JuJGd39kuxcRmfCivMCNuz8KPFrQdk/e9J8Ms+3fAX8XXXXHpXQaSkRkWGW/wD0e6DSUiMjwFBZAMqGehYjIcBQWQDJmGqJcRGQYCgt066yISCkKCyCZMF2zEBEZhsICSMT0pjwRkeEoLICULnCLiAxLYUEwNpROQ4mIDE1hwcAQ5U4up8AQESlGYUEQFoCuW4iIDEFhAaQTwR9Db0ZhISJSjMICqEgGb2zt1WCCIiJFKSw4HhYaeVZEpDiFBVAZhkVPv05DiYgUo7AAKpLBH0OPehYiIkUpLDh+GkphISJSnMKCvJ6F7oYSESlKYQGkE+pZiIgMR2GBTkOJiJSisOD4aahe3Q0lIlKUwoK8nkVGPQsRkWIUFuQ/Z6GwEBEpRmFB/jULnYYSESlGYQHEY0YybupZiIgMQWERqkjE1bMQERmCwiKUTsY1kKCIyBAUFqGKZExDlIuIDEFhEapOJejqU1iIiBSjsAhVpeN09mXKXYaIyLiksAhVpxJ09iosRESKiTQszOw6M9tiZq1mdleR5Xea2SYz22BmPzaz+XnLsma2PvysjrJOgOp0XKehRESGkIhqx2YWB+4D3gPsBNaa2Wp335S32gtAs7t3mdkfAp8DbgmXdbv78qjqK1SdSug0lIjIEKLsWawEWt19m7v3AQ8DN+Wv4O5PuXtXOPssMCfCeoZVlY7T1auehYhIMVGGxWxgR978zrBtKB8D/jtvvsLMWszsWTP7QLENzOz2cJ2W9vb2Myq2OpXgmK5ZiIgUFdlpqFNhZh8BmoFfzWue7+5tZnYu8KSZveTur+Vv5+4PAA8ANDc3+5nUUJ1O0JvJkcnmSMR13V9EJF+Ufyu2AXPz5ueEbScws3cDnwRudPfegXZ3bwu/twFPA5dGWCtVqWAwwS49mCcicpIow2ItsNjMFppZClgFnHBXk5ldCtxPEBT78tobzCwdTjcCVwL5F8ZHXXU66GTpuoWIyMkiOw3l7hkzuwN4DIgDD7r7RjO7F2hx99XA54Ea4NtmBvCmu98IvAW438xyBIH2mYK7qEbdQM9C1y1ERE4W6TULd38UeLSg7Z686XcPsd0vgIujrK1QdSrsWej2WRGRk+hKbmjgNNSxHoWFiEghhUWorjIIi6M9/WWuRERk/FFYhOorkwAc6VZYiIgUUliEFBYiIkNTWIRq0gniMVNYiIgUobAImRl1FQmFhYhIEQqLPPWVSY50624oEZFCCos8QVioZyEiUkhhkadOYSEiUpTCIk99ZZKjCgsRkZMoLPLoNJSISHEKizwDYeF+Rq/GEBGZdBQWeeork2RzTlefhikXEcmnsMijp7hFRIpTWORRWIiIFKewyKOwEBEpTmGRp05hISJSlMIij3oWIiLFKSzyDPYsuhQWIiL5FBZ56ioSJOPGgc6+cpciIjKuKCzymBlNNWnaO3rLXYqIyLiisCjQWJtm/zGFhYhIPoVFAfUsREROprAo0FSbpl09CxGREygsCjTVpjlwrJdsToMJiogMUFgUaKxJk3M41KU7okREBigsCjTVpgF03UJEJI/CooDCQkTkZAqLAk01CgsRkUKRhoWZXWdmW8ys1czuKrL8TjPbZGYbzOzHZjY/b9ltZvZq+LktyjrzNYY9Cz1rISJyXGRhYWZx4D7gemAJcKuZLSlY7QWg2d0vAb4DfC7cdjrwKeAKYCXwKTNriKrWfNWpOJXJuHoWIiJ5ouxZrARa3X2bu/cBDwM35a/g7k+5e1c4+ywwJ5x+L/C4ux9090PA48B1EdY6yMxoqk2zT2EhIjIoyrCYDezIm98Ztg3lY8B/n+a2o+rsugr2HO0Zqx8nIjLujSgszOw/RtJ2uszsI0Az8PlT3O52M2sxs5b29vbRKodZ0ypoO9Q9avsTEZnoRtqzuCh/JrwecXmJbdqAuXnzc8K2E5jZu4FPAje6e++pbOvuD7h7s7s3NzU1lTyIkZrdUMmeoz1ksrlR26eIyEQ2bFiY2d1m1gFcYmZHw08HsA94pMS+1wKLzWyhmaWAVcDqgv1fCtxPEBT78hY9BlxrZg3hhe1rw7YxMXtaFdmcs1fXLUREgBJh4e7/x91rgc+7e134qXX3Ge5+d4ltM8AdBH/Jbwa+5e4bzexeM7sxXO3zQA3wbTNbb2arw20PAp8mCJy1wL1h25iY3VAJoFNRIiKhxAjX+4GZVbt7Z3h94TLgS+6+fbiN3P1R4NGCtnvypt89zLYPAg+OsL5RNXtaEBa7DissRERg5Ncs/gXoMrNlwJ8BrwFfj6yqMps1rQKANoWFiAgw8rDIuLsTPCfxFXe/D6iNrqzyqkolmF6dYqdOQ4mIACM/DdVhZncDvw38ipnFgGR0ZZXf7GmV6lmIiIRG2rO4BegFPuruewhuZT2lZyImmtnTKmk71FV6RRGRKWBEYREGxENAvZndAPS4+6S9ZgEwd3olOw51k9Mb80RERvwE928Ca4APAb8JPGdmN0dZWLktbKyhL5Nj1xGdihIRGek1i08CKwYenDOzJuAJgpFiJ6WFjdUAvL6/kzkNVWWuRkSkvEZ6zSJW8IT1gVPYdkI6t+l4WIiITHUj7Vn80MweA74Zzt9CwcN2k83M2jTVqTjb2hUWIiLDhoWZLQLOcve/MLNfB64KF/2S4IL3pGVmLGyqVs9CRITSp5L+CTgK4O7/5e53uvudwPfCZZPawsYahYWICKXD4ix3f6mwMWxbEElF48jCxmp2HuqiN5MtdykiImVVKiymDbOscjQLGY/Oa6om57rILSJSKixazOz3ChvN7H8C66Ipafy44Oxg+KstezrKXImISHmVuhvqT4HvmdlvcTwcmoEU8MEoCxsPzm2sIRk3XtnTwU3lLkZEpIyGDQt33wu83czeCSwNm/+fuz8ZeWXjQCoR47ymGl7ZfbTcpYiIlNWInrNw96eApyKuZVx6yzl1PLvtQLnLEBEpq0n9FPZouODsWnYf6eFwV1+5SxERKRuFRQkXhhe5X9FFbhGZwhQWJSyZVQfAy21HylyJiEj5KCxKmFlbwaz6Cl7cqbAQkalLYTECy+ZOY8POw+UuQ0SkbBQWI7Bs7jS2H+jiUKcucovI1KSwGIFlc4JRT15U70JEpiiFxQhcPKceM1i/Q2EhIlOTwmIEatIJLjirlnXbD5W7FBGRslBYjNAVC6ezbvsh+rO5cpciIjLmFBYjtHLhDLr6smzcpXGiRGTqUViM0IqFDQCseV3jRInI1KOwGKGZtRWc21TNc9sOlrsUEZExF2lYmNl1ZrbFzFrN7K4iy682s+fNLGNmNxcsy5rZ+vCzOso6R+rt583g2W0H6MvouoWITC2RhYWZxYH7gOuBJcCtZrakYLU3gd8BvlFkF93uvjz83BhVnafiV8+fSWdflpY31LsQkaklyp7FSqDV3be5ex/wMJz4wjl3f8PdNwAT4p/qbz9vBql4jKe3tpe7FBGRMRVlWMwGduTN7wzbRqrCzFrM7Fkz+8DolnZ6qtMJVi6cztNb9pW7FBGRMTWeL3DPd/dm4MPAP5nZeYUrmNntYaC0tLePzb/233FBE1v3HmPX4e4x+XkiIuNBlGHRBszNm58Tto2Iu7eF39uAp4FLi6zzgLs3u3tzU1PTmVU7Qu+4IPg5T2/RqSgRmTqiDIu1wGIzW2hmKWAVMKK7msyswczS4XQjcCWwKbJKT8F5TTXMnV7J45v2lLsUEZExE1lYuHsGuAN4DNgMfMvdN5rZvWZ2I4CZrTCzncCHgPvNbGO4+VuAFjN7EXgK+Iy7j4uwMDPet/Qcfvrqfr2XW0SmjESUO3f3R4FHC9ruyZteS3B6qnC7XwAXR1nbmbjhklnc/8w2Htu4h1tWzCt3OSIikRvPF7jHraWz65g/o4ofbNhd7lJERMaEwuI0mBk3XHIOP2/dz/5jveUuR0QkcgqL03TDJbPIOTz6knoXIjL5KSxO04Vn17LknDq+uWYH7l7uckREIqWwOE1mxq1XzGPz7qNs2Hmk3OWIiERKYXEGblo+i8pknG+uebPcpYiIREphcQbqKpL82rJzWP3iLo71ZspdjohIZBQWZ+jDV8ynqy/Lt9buKL2yiMgEpbA4Q8vnTmPFgga+9rPX6c9OiJHWRUROmcJiFPz+1efRdrhbt9GKyKSlsBgF11w4k0Uza/jqT7bpNloRmZQUFqMgFjN+/+pz2bz7KI9v2lvuckRERp3CYpR88NLZnNtYzT/+aCvZnHoXIjK5KCxGSSIe485rz2fL3g5WvzjidzyJiEwICotR9L6l57DknDq+8PhWevqz5S5HRGTUKCxGUSxmfPL9b2HHwW7+9Zlt5S5HRGTUKCxG2ZWLGnn/xedw39Ot7DzUVe5yRERGhcIiAn/9/rdgGH+zepNupRWRSUFhEYHZ0yr5xHsW88TmvXx/vS52i8jEp7CIyMeuOpfm+Q3c88hGdh/pLnc5IiJnRGERkXjM+IcPLSOTdf7quy/pdJSITGgKiwgtaKzmr993Ic9sbedff6q7o0Rk4lJYROwjb53P9UvP5rM/3MIvXztQ7nJERE6LwiJiZsbnP7SMBTOquOMbz+v6hYhMSAqLMVCTTnD/b19OT3+W3/t6i96qJyITjsJijCyaWctXPnwZm3d38EcPPa8XJYnIhKKwGEPvvHAmf//BpTyztZ27vvsSOY1OKyITRKLcBUw1t6yYx54jvXzxia2kEjH+7gNLicWs3GWJiAxLYVEGf/yuRfRls9z31Gu4O3//wYsVGCIyriksysDM+PNrLyBmxpefbKWnP8tnb76EdCJe7tJERIqK9JqFmV1nZlvMrNXM7iqy/Goze97MMmZ2c8Gy28zs1fBzW5R1loOZced7zucv3nsB31+/i9seXMORrv5ylyUiUlRkYWFmceA+4HpgCXCrmS0pWO1N4HeAbxRsOx34FHAFsBL4lJk1RFVruZgZH3/nIr60ajnrth/iN776C3Yc1LDmIjL+RNmzWAm0uvs2d+8DHgZuyl/B3d9w9w1A4X2k7wUed/eD7n4IeBy4LsJay+qm5bP5+kevYN/RHm748s946pV95S5JROQEUYbFbGBH3vzOsG3UtjWz282sxcxa2tvbT7vQ8eBt581g9R1XMWtaJb/7b2v5h8e2kNWttSIyTkzo5yzc/QF3b3b35qampnKXc8YWNFbzvT96O7c0z+UrT7Wy6oFf8sb+znKXJSISaVi0AXPz5ueEbVFvO6FVJON89uZL+OIty3hlTwfXf+mn/Psv3tADfCJSVlGGxVpgsZktNLMUsApYPcJtHwOuNbOG8ML2tWHblPHBS+fwo09czYqF0/nU6o2seuBZNu8+Wu6yRGSKiiws3D0D3EHwl/xm4FvuvtHM7jWzGwHMbIWZ7QQ+BNxvZhvDbQ8CnyYInLXAvWHblHJOfSX//rsr+NxvXMKr+zq44cs/429Wb+RIt26xFZGxZZPlDW7Nzc3e0tJS7jIic7irj3/80VYeem4706tT/PG7FrNqxTxSiQl92UlEyszM1rl7c6n19DfNBDGtKsWnP7CU1XdcxblNNdzzyEau+cen+XbLDjIawVZEIqawmGCWzq7nP29/K1//6EoaqlL8xXc2cO0Xn+HhNW/S058td3kiMknpNNQE5u48tnEPX36ylY27jtJYk+Z3r1zAR66YT31VstzlicgEMNLTUAqLScDd+cVrB7j/mW08s7WdqlScm5bP4sMr53PxnPpylyci49hIw0Kjzk4CZsaVixq5clEjm3cf5cGfvc73Xmjjm2t2sHR2HbeunMevLZtFXYV6GyJyetSzmKSO9vTzyAttPPTcm7yyp4NUIsa7LpzJTctn8Y4LZlKR1HDoIqLTUBJyd17ceYTvv9DGDzbsZv+xXmrTCd679GyuX3o2Vy5qVHCITGEKCzlJJpvj2W0HeWR9Gz98eQ8dvRkqkjGuWtTEe5bM5JoLz6KpNl3uMkVkDCksZFh9mRzPvX6AJzbt5YnN+2g73I0ZXDK7fvD6x+XzG9TrEJnkFBYyYu7OK3s6eGLTXn6ytZ31Ow6TyTmpRIwVCxp4+3mNvO28GSydVa8nxkUmGYWFnLZjvRnWvH6An7ce4Oet+3llTwcA6USMS+bUc9m8Bi6b38Bl8xp02kpkglNYyKjZf6yXNa8f5Pnth3j+zUO83HaUvnCIkXnTq1g2dxpLZ9Vx0ax6LppVR0N1qswVi8hIKSwkMj39WTbuOsLz2w+zbvshXmo7Qtvh7sHls6dVclEYHktm1XH+WTXMaagiHrMyVi0ixeihPIlMRTLO5fOnc/n86fxe2Haos4+Nu46ycdcRXg6/H9+8l4F/i6QTMc5rquH8s2pYfFYti2bWsHhmDfOmV5GI6zqIyHinsJBR0VCd4qrFjVy1uHGwrbM3wyt7Omjd18Gre4/x6r5jrH3jEN9fv2twnWTcmNtQxfwZVcyfUc2C8Hv+jCrmNFTpgrrIOKGwkMhUpxNcPr+By+c3nNB+rDfDa/uC8Gjdd4w3D3byxv4u1rx+kM6+4yPnxgxmN1Qyf3o1s6dVMmtaJbOmVQxOn11foVt7RcaIwkLGXE06wbK501g2d9oJ7e7O/mN9bD/QyfYDXWw/0MkbB7rYfrCLJ7fso72j96R9NdakmT2tIgySSs6qS3NWXQVNtWlm1lYwsy5NbTqBma6XiJwJhYWMG2ZGU22apto0zQumn7S8N5Nlz5Ee2g53s+twD7sOd7PrcDdth7vZureDp7e0013knR4VyVgQHLV5QVKXpqkmzYyaFNOr08yoTjG9OkVVKq5gESlCYSETRjoRD69nVBdd7u509GbYd7SHfUd72dfRy76O49N7j/awefdRfrK1l2O9maL7SCVig8GR/5lRnaIh/K6vTFFfmaS+Kkl9ZZJqBYxMAQoLmTTMjLqKJHUVSRbNrB123a6+DO0dvRzo7ONQZx8HOvs4WPA50NnH9gNdHOzsGzJcABIxo64yCI6B7+CTyJsOl1ckqalIUJ1OUJtOUFORoDKpsJHxT2EhU1JVKsH8GYkheymFevqzHO7q50BnL0e6+jnSPcynq483D3RypLufoz0Zsrnhn2WKGSeER006DJP86cFlSarTcapTCapScSpTcapSQeAE03Eqk3FieqZFRpnCQmQEKpJxzq6Pc3Z9xSlt5+4c680MBsmxngzHeoNPR0+GziLTA/N7jvQE8z0ZjvVlOJXnZyuSscEQqRoIkYFgScWpSg60JQaXpxMx0snguyIZpyJveqhvPWg5dSgsRCJkZtRWJKmtSDKnofT6Q8nlnO7+7GCQdPdl6erL0NWfDaezdPdl6BqY7g+X9+Uvz7Kvo2dwOlgnS18md9p1JeNGOhGnIhkjnYiTDr+D+TB0BttjpBIxUvE4yYSRjsdIxsO2RN50/MT54Gcc3y4Vbpc+YZ1gPZ3Oi47CQmQCiMWM6vCU1Fl1o7vvTDZHd3+W3kyOnrzvnv4cvZksveF3T5HvniLb5X8f6uw7Yf3+bI6+TI6+bPAZ7dGGTgwbOx4ksRiJuJGIx0jGLJgeaIsF656wbHA6WCcZixGP2eB6iZiRzFtWbJtE7MR952+TiBnxcHnwbcQL2mPGuAo/hYXIFJeIx6iNxxj+loDR5+5kcx4Ex0CAZHL0Zz2YP6Ht+Hx/NkdvXtvgskyOvnDbwlDKZHNksk5/zgenOzPB9aT+bNiWc/rDZZlcbrB9YJsSl54icTw8wu/48XCJWRBs8Zhx0ax6vnzrpdHWEuneRUSGYHb8X/FVE2Cg4lzOyeRODJLiAeP05463FS7L5HJkw30Nfof7Oqk9F7Znh2gP5+c2VEZ+/AoLEZERiMWMVMxIMTXHK5uaRy0iIqdEYSEiIiVFGhZmdp2ZbTGzVjO7q8jytJn9Z7j8OTNbELYvMLNuM1sffr4aZZ0iIjK8yK5ZmFkcuA94D7ATWGtmq919U95qHwMOufsiM1sFfBa4JVz2mrsvj6o+EREZuSh7FiuBVnff5u59wMPATQXr3AT8ezj9HeBdNp5uLBYRESDasJgN7Mib3xm2FV3H3TPAEWBGuGyhmb1gZj8xs1+JsE4RESlhvN46uxuY5+4HzOxy4PtmdpG7H81fycxuB24HmDdvXhnKFBGZGqLsWbQBc/Pm54RtRdcxswRQDxxw9153PwDg7uuA14DzC3+Auz/g7s3u3tzU1BTBIYiICETbs1gLLDazhQShsAr4cME6q4HbgF8CNwNPurubWRNw0N2zZnYusBjYNtwPW7du3X4z234G9TYC+89g+4lIxzw16JinhtM95vkjWSmysHD3jJndATwGxIEH3X2jmd0LtLj7auBrwH+YWStwkCBQAK4G7jWzfiAH/IG7Hyzx886oa2FmLe7efCb7mGh0zFODjnlqiPqYI71m4e6PAo8WtN2TN90DfKjIdt8FvhtlbSIiMnJ6gltEREpSWBz3QLkLKAMd89SgY54aIj1m89F++4iIiEw66lmIiEhJUz4sSg12OFGZ2YNmts/MXs5rm25mj5vZq+F3Q9huZvbP4Z/BBjO7rHyVnz4zm2tmT5nZJjPbaGZ/ErZP2uM2swozW2NmL4bH/Ldh+8JwcM7WcLDOVNhedPDOicjM4uEoDz8I5yf1MZvZG2b2Uji4akvYNma/21M6LPIGO7weWALcamZLylvVqPk34LqCtruAH7v7YuDH4TwEx784/NwO/MsY1TjaMsCfufsS4K3Ax8P/npP5uHuBa9x9GbAcuM7M3kowKOcX3X0RcIhg0E7IG7wT+GK43kT1J8DmvPmpcMzvdPflebfIjt3vtrtP2Q/wNuCxvPm7gbvLXdcoHt8C4OW8+S3AOeH0OcCWcPp+4NZi603kD/AIwajHU+K4gSrgeeAKgoezEmH74O85wXNPbwunE+F6Vu7aT+NY54R/OV4D/ACwKXDMbwCNBW1j9rs9pXsWjGyww8nkLHffHU7vAc4Kpyfdn0N4quFS4Dkm+XGHp2PWA/uAxwmGxznsweCccOJxDTd450TyT8BfEjy0C8ExTPZjduBHZrYuHBcPxvB3e7wOJCgRc3c3s0l5K5yZ1RA81Pmn7n40f9T7yXjc7p4FlpvZNOB7wIVlLilSZnYDsM/d15nZO8pdzxi6yt3bzGwm8LiZvZK/MOrf7anesxjJYIeTyV4zOwcg/N4Xtk+aPwczSxIExUPu/l9h86Q/bgB3Pww8RXAKZlo4OCeceFxFB+8c41LP1JXAjWb2BsF7cq4BvsTkPmbcvS383kfwj4KVjOHv9lQPi8HBDsM7J1YRDG44WQ0M3Ej4/Uhe+/8I76B4K3Akr2s7YVjQhfgasNndv5C3aNIet5k1hT0KzKyS4BrNZoLQuDlcrfCYB/4sBgfvHLuKz5y73+3uc9x9AcH/s0+6+28xiY/ZzKrNrHZgGrgWeJmx/N0u90Wbcn+A9wFbCc7zfrLc9YzicX2T4L0g/QTnKz9GcJ72x8CrwBPA9HBdI7gr7DXgJaC53PWf5jFfRXBedwOwPvy8bzIfN3AJ8EJ4zC8D94Tt5wJrgFbg20A6bK8I51vD5eeW+xjO8PjfAfxgsh9zeGwvhp+NA39XjeXvtp7gFhGRkqb6aSgRERkBhYWIiJSksBARkZIUFiIiUpLCQkRESlJYiITM7Fj4vcDMPjzK+/7rgvlfjOb+RaKmsBA52QLglMIi78nhoZwQFu7+9lOsSaSsFBYiJ/sM8CvhewM+EQ7U93kzWxu+G+D3AczsHWb2UzNbDWwK274fDvS2cWCwNzP7DFAZ7u+hsG2gF2Phvl8O31VwS96+nzaz75jZK2b2UPiEOmb2GQve2bHBzP5hzP90ZErSQIIiJ7sL+HN3vwEg/Ev/iLuvMLM08HMz+1G47mXAUnd/PZz/qLsfDIfeWGtm33X3u8zsDndfXuRn/TrBeyiWAY1vT3eOAAABdElEQVThNs+Eyy4FLgJ2AT8HrjSzzcAHgQvd3QeG+hCJmnoWIqVdSzDOznqCIc9nELxUBmBNXlAA/LGZvQg8SzCQ22KGdxXwTXfPuvte4CfAirx973T3HMHQJQsIhtfuAb5mZr8OdJ3x0YmMgMJCpDQD/pcHbyhb7u4L3X2gZ9E5uFIwXPa7CV60s4xgzKaKM/i5vXnTWYIX+2QIRhv9DnAD8MMz2L/IiCksRE7WAdTmzT8G/GE4/Dlmdn448meheoLXd3aZ2YUEr3Yd0D+wfYGfAreE10WagKsJBrsrKnxXR727Pwp8guD0lUjkdM1C5GQbgGx4OunfCN6VsAB4PrzI3A58oMh2PwT+ILyusIXgVNSAB4ANZva8B8NpD/gewfsnXiQYMfcv3X1PGDbF1AKPmFkFQY/nztM7RJFTo1FnRUSkJJ2GEhGRkhQWIiJSksJCRERKUliIiEhJCgsRESlJYSEiIiUpLEREpCSFhYiIlPT/AUBBXGNPuaCCAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1133d8978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import SoftmaxRegression\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "lr = SoftmaxRegression(eta=0.01, \n",
    "                       epochs=500, \n",
    "                       minibatches=1, \n",
    "                       random_seed=1,\n",
    "                       print_progress=3)\n",
    "lr.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=lr)\n",
    "plt.title('Softmax Regression - Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(lr.cost_)), lr.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:54:23.305938Z",
     "start_time": "2018-04-19T10:54:23.300207Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Last 3 Class Labels: [2 2 2]\n"
     ]
    }
   ],
   "source": [
    "y_pred = lr.predict(X)\n",
    "print('Last 3 Class Labels: %s' % y_pred[-3:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:54:36.572737Z",
     "start_time": "2018-04-19T10:54:36.566744Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Last 3 Class Labels:\n",
      " [[9.18728149e-09 1.68894679e-02 9.83110523e-01]\n",
      " [2.97052325e-11 7.26356627e-04 9.99273643e-01]\n",
      " [1.57464093e-06 1.57779528e-01 8.42218897e-01]]\n"
     ]
    }
   ],
   "source": [
    "y_pred = lr.predict_proba(X)\n",
    "print('Last 3 Class Labels:\\n %s' % y_pred[-3:])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Stochastic Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:55:09.680547Z",
     "start_time": "2018-04-19T10:55:07.341337Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x122631be0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xt4FPd97/H3d3e1qxtCAsTFAgzY+ILtGDsCOxeTJvE1F+OkbmynSX2aNK7T+CSxm3PqJH2cxn2SJmnrpk2dOKT1aZvGcS5uEk4O8aXxPYlthA3GYAziYm4GBAiE7lrpe/6YESyLpJVAo11Jn9fz7LMzv5lZfYcFfZj5zfzG3B0REZGBxPJdgIiIFD6FhYiI5KSwEBGRnBQWIiKSk8JCRERyUliIiEhOCgsREclJYSEiIjkpLEREJKdEvgsYLlOmTPE5c+bkuwwRkVFl1apV+929Otd6YyYs5syZQ11dXb7LEBEZVczs9cGsp9NQIiKSk8JCRERyUliIiEhOCgsREclJYSEiIjkpLEREJCeFhYiI5DTuw6KlI809j23kpe2N+S5FRKRgjfuw6Ej38E+/3sSaHYfyXYqISMEa92GRTAR/BJ3dPXmuRESkcCks4mFYpBUWIiL9GfdhURQ3QGEhIjKQSMPCzK42s9fMrN7M7uxj+a1mttbMVpvZs2a2IGyfY2ZtYftqM7svwhpJJmJ06DSUiEi/Iht11sziwL3AFcBOYKWZLXf39RmrPeDu94XrXwvcA1wdLtvs7gujqi9TMh7TkYWIyACiPLJYDNS7+xZ37wQeBJZmruDuTRmzZYBHWE+/kgmFhYjIQKIMixpgR8b8zrDtOGb2KTPbDHwD+HTGorlm9pKZPWVml0VYp44sRERyyHsHt7vf6+5nAH8B/GXY/AYw290vAu4AHjCziuxtzewWM6szs7qGhoaTriGZiOnSWRGRAUQZFruAWRnzM8O2/jwIXAfg7h3ufiCcXgVsBs7K3sDdl7l7rbvXVlfnfCpgv5KJGF0KCxGRfkUZFiuB+WY218ySwI3A8swVzGx+xux7gU1he3XYQY6ZzQPmA1uiKlSnoUREBhbZ1VDunjaz24BHgDhwv7uvM7O7gTp3Xw7cZmaXA11AI3BzuPkS4G4z6wJ6gFvd/WBUtSYTMToUFiIi/YosLADcfQWwIqvtrozpz/Sz3UPAQ1HWlklXQ4mIDCzvHdyFIKUObhGRASksUJ+FiEguCgt0GkpEJBeFBbrPQkQkF4UFOg0lIpKLwgKdhhIRyUVhARTpyEJEZEAKC4JLZ/U8CxGR/iksOHYayj0vI6SLiBQ8hQXHnsPd1a2wEBHpi8KC4MgC0OWzIiL9UFhwLCy61MktItInhQU6shARyUVhwbE+C10+KyLSN4UFx44s9EwLEZG+KSwI7rMAHVmIiPRHYYH6LEREclFYAMl4HNCRhYhIfxQWZBxZKCxERPqksCDzNFR3nisRESlMCgugKG6AjixERPqjsODY1VC6dFZEpG8KC9TBLSKSi8ICXTorIpKLwgJdDSUikkukYWFmV5vZa2ZWb2Z39rH8VjNba2arzexZM1uQsezz4XavmdlVUdZZmgxOQ7V26mooEZG+RBYWZhYH7gWuARYAN2WGQegBd7/A3RcC3wDuCbddANwInAdcDXw7/LxIpBIx4jGjpSMd1Y8QERnVojyyWAzUu/sWd+8EHgSWZq7g7k0Zs2VA76PqlgIPunuHu28F6sPPi4SZUZ5KKCxERPqRiPCza4AdGfM7gUuyVzKzTwF3AEngXRnbPpe1bU0f294C3AIwe/bsUyq2PJWguUOnoURE+pL3Dm53v9fdzwD+AvjLIW67zN1r3b22urr6lOooS8Vp7ug6pc8QERmrogyLXcCsjPmZYVt/HgSuO8ltT1lZKkGLjixERPoUZVisBOab2VwzSxJ0WC/PXMHM5mfMvhfYFE4vB240s5SZzQXmAy9EWGt4Gkp9FiIifYmsz8Ld02Z2G/AIEAfud/d1ZnY3UOfuy4HbzOxyoAtoBG4Ot11nZj8G1gNp4FPuHul/+8tTCfYcbo/yR4iIjFpRdnDj7iuAFVltd2VMf2aAbb8CfCW66o5XpiMLEZF+5b2Du1DoNJSISP8UFqHe+yzcPffKIiLjjMIiVJZK0OPQ3qXxoUREsiksQuWpYDSRI7rXQkTkBAqLUHlx0Nevey1ERE6ksAiVJXvDQp3cIiLZFBah8lQQFroiSkTkRAqLUO9pqOZ2hYWISDaFRagsPLJo6VRYiIhkU1iEdBpKRKR/CotQb1gc0WkoEZETKCxCpck4yXiMQ626z0JEJJvCImRmVJYWcai1M9+liIgUHIVFhkllSQ62KCxERLIpLDIERxY6DSUikk1hkWFSWZKDOg0lInIChUWGytKk+ixERPqgsMhQVVpEY2uXnmkhIpJFYZGhqjRJd4/TpHstRESOo7DIUFWaBKBRV0SJiBxHYZFhUlkYFuq3EBE5jsIiQ2VpEYAunxURyaKwyNB7ZKEb80REjqewyFBZqtNQIiJ9iTQszOxqM3vNzOrN7M4+lt9hZuvN7GUz+7WZnZ6xrNvMVoev5VHW2auiOEEyHqOhuWMkfpyIyKiRiOqDzSwO3AtcAewEVprZcndfn7HaS0Ctu7ea2SeBbwA3hMva3H1hVPX1UzNTK1I0NCksREQyRXlksRiod/ct7t4JPAgszVzB3Z9w99Zw9jlgZoT1DMrUCSn2HmnPdxkiIgUlyrCoAXZkzO8M2/rzceBXGfPFZlZnZs+Z2XVRFNiXaRXF7NWRhYjIcSI7DTUUZvYRoBZ4R0bz6e6+y8zmAY+b2Vp335y13S3ALQCzZ88ellqmVRTzm/r9w/JZIiJjRZRHFruAWRnzM8O245jZ5cAXgWvd/eh/6d19V/i+BXgSuCh7W3df5u617l5bXV09LEVXT0jR1J6mrbN7WD5PRGQsiDIsVgLzzWyumSWBG4Hjrmoys4uA7xIExb6M9iozS4XTU4C3AZkd45GZVlEMwD71W4iIHBVZWLh7GrgNeAR4Ffixu68zs7vN7Npwtb8FyoGfZF0iey5QZ2ZrgCeAr2VdRRWZaRUpAPYdUb+FiEivSPss3H0FsCKr7a6M6cv72e63wAVR1tafqROCI4u9TTqyEBHppTu4s/QeWeiKKBGRYxQWWSaWFFFcFGPP4bZ8lyIiUjAUFlnMjJrKEnY2KixERHopLPows6pUYSEikkFh0YeZVSXsbGzNvaKIyDihsOjDzKpSGlu7aOnQs7hFREBh0aeZVSUA7DqkU1EiIqCw6FNNGBY6FSUiElBY9GHm0bDQkYWICCgs+lRdnqK4KMbrB3RkISICgwwLM/v+YNrGCjNjzuQytu1vyXcpIiIFYbBHFudlzoSPTH3z8JdTOOZVl7FVYSEiAuQICzP7vJkdAd5kZk3h6wiwD/jFiFSYJ3OnlLH9YCtd3T35LkVEJO8GDAt3/xt3nwD8rbtXhK8J7j7Z3T8/QjXmxdwp5aR7XJ3cIiIM/jTUL82sDIJHoJrZPWZ2eoR15d3cKWUAbN3fnOdKRETyb7Bh8R2g1cwuBP4c2Az8R2RVFYB5YVhsaVC/hYjIYMMi7e4OLAX+2d3vBSZEV1b+VZUlmVSWZHODjixERAb7pLwjZvZ54KPAZWYWA4qiK6swnDWtnA17juS7DBGRvBvskcUNQAfwMXffA8wkeH72mHb2tAls3HOE4KBKRGT8GlRYhAHxA2Cimb0PaHf3Md1nAXDW9Am0dHbriigRGfcGewf3h4AXgD8APgQ8b2bXR1lYIThnetAts3GvTkWJyPg22D6LLwKL3H0fgJlVA/8N/DSqwgrB/GlBWGzYc4R3nzstz9WIiOTPYPssYr1BETowhG1HrYriImZWlbB+d1O+SxERyavBHlk8bGaPAD8M528AVkRTUmG5oGYir+w+nO8yRETyKtfYUGea2dvc/X8B3wXeFL5+ByzL9eFmdrWZvWZm9WZ2Zx/L7zCz9Wb2spn9OvOucDO72cw2ha+bh7xnw+T8mom8fqCVw21d+SpBRCTvcp1K+ibQBODu/+Xud7j7HcDPwmX9CkemvRe4BlgA3GRmC7JWewmodfc3EfR/fCPcdhLwJeASYDHwJTOrGsqODZcLaiYCsG6Xji5EZPzKFRbT3H1tdmPYNifHtouBenff4u6dwIMEd4Bnfs4T7t77hKHnCO7fALgKeMzdD7p7I/AYcHWOnxeJ88OwWKuwEJFxLFdYVA6wrCTHtjXAjoz5nWFbfz4O/Ookt43MpLIksyaV8NL2Q/n48SIiBSFXWNSZ2SeyG83sT4BVw1WEmX0EqGWId4Wb2S1mVmdmdQ0NDcNVzgkWzZnEym0HdSe3iIxbucLis8Afm9mTZvb34espgqOAz+TYdhcwK2N+Zth2HDO7nOA+jmvdvWMo27r7Mnevdffa6urqHOWcvEVzJnGgpZMtenKeiIxTuR5+tNfd3wp8GdgWvr7s7m8JhwAZyEpgvpnNNbMkcCOwPHMFM7uI4Cqra7Pu43gEuNLMqsKO7SvDtrxYNGcSAHXbDuarBBGRvBrUfRbu/gTwxFA+2N3TZnYbwS/5OHC/u68zs7uBOndfTnDaqRz4iZkBbHf3a939oJn9NUHgANzt7nn7TX1GdRmTypK8sLWRGxbNzlcZIiJ5M9ib8k6Ku68g6+Y9d78rY/ryAba9H7g/uuoGz8yoPb2KlTqyEJFxaswP2TFcFs+dxPaDrextas93KSIiI05hMUi9/RYvbNXRhYiMPwqLQVpwWgWlyTjPbz2Q71JEREacwmKQiuIx3jJvMk9tbND9FiIy7igshuCd50xlx8E2Njc057sUEZERpbAYgneeMxWAxzfsy7GmiMjYorAYgprKEs6eNoEnNkQ3tIiISCFSWAzRO8+ZysptB2lq1/MtRGT8UFgM0bvOmUq6x3l20/58lyIiMmIUFkN08exKJpYU8d+v7s13KSIiI0ZhMUSJeIwrF0zj0XV7ae/qznc5IiIjQmFxEpYurKG5I80TuipKRMYJhcVJeMsZk5lSnuIXq3fnuxQRkRGhsDgJ8Zjx/gtn8Phr+zjcpquiRGTsU1icpKULa+hM9/DIulzPgBIRGf0UFifpwpkTmTO5lIdW7cx3KSIikVNYnCQz44ZFs3l+60Hq9x3JdzkiIpFSWJyCD9XOpChu/Odz2/NdiohIpBQWp2ByeYprzp/BQy/upK1T91yIyNilsDhFH7n0dI60p/m/a3QZrYiMXQqLU7RoThVnT5vA/b/ZqociiciYpbA4RWbGJ5bMY8OeIzz5moYuF5GxSWExDJYuPI2ayhK+/WR9vksREYmEwmIYFMVjfOKyuazc1sgLWw/muxwRkWEXaViY2dVm9pqZ1ZvZnX0sX2JmL5pZ2syuz1rWbWarw9fyKOscDjcsms3ksiTfenxTvksRERl2kYWFmcWBe4FrgAXATWa2IGu17cD/AB7o4yPa3H1h+Lo2qjqHS0kyzq3vOINnNu3nN/V6MJKIjC1RHlksBurdfYu7dwIPAkszV3D3be7+MtATYR0j5qNvOZ3TJhbztV9toKdHV0aJyNgRZVjUADsy5neGbYNVbGZ1ZvacmV03vKVFo7gozu1XnMXaXYdZ8cob+S5HRGTYFHIH9+nuXgt8GPimmZ2RvYKZ3RIGSl1DQ2FctvrBi2dy9rQJfP3hDXqSnoiMGVGGxS5gVsb8zLBtUNx9V/i+BXgSuKiPdZa5e62711ZXV59atcMkHjO+9P4F7DjYxref0KW0IjI2RBkWK4H5ZjbXzJLAjcCgrmoysyozS4XTU4C3Aesjq3SYvfXMKVy38DTue2oLWxqa812OiMgpiyws3D0N3AY8ArwK/Njd15nZ3WZ2LYCZLTKzncAfAN81s3Xh5ucCdWa2BngC+Jq7j5qwAPjCe88lVRTjL3/+ioYBEZFRz8bKL7La2lqvq6vLdxnHeeD57XzhZ2v56+vO56OXnp7vckRETmBmq8L+4QEVcgf3qHfT4llcNn8Kf7PiVbYfaM13OSIiJ01hESEz4+u//ybiZvz5T1aT7h4Tt5OIyDiksIjYaZUl/PV157NyWyP3PLYx3+WIiJwUhcUIuO6iGm5aPItvP7mZJzbsy3c5IiJDprAYIV96/3mcO6OCz/5oNVv3t+S7HBGRIVFYjJDiojjf/cibiRl8/N9XcritK98liYgMmsJiBM2eXMp3PvJmth9o5bYHXqQzrQ5vERkdFBYj7NJ5k/nqBy/gmU37uePHq+nW6LQiMgok8l3AePSh2lkcau3kqys2MKE4wVc/cAFmlu+yRET6pbDIk1uWnMHhti7ufWIz5akEX3jPuQoMESlYCos8+tyVZ3OkPc33ntlKV7dz1/sWEIspMESk8Cgs8sjM+PK155GMx/iXZ7dypD3N13//AhJxdSWJSGFRWOSZmfHF955LRUkR9zy2keaOLr55w0WUJOP5Lk1E5Cj9F7YAmBmffvd8vvT+BTy6fi83LPsde5va812WiMhRCosC8sdvm8uyj9ZSv6+Za//5WdbuPJzvkkREAIVFwbliwTQe+uRbScRiXH/fb/nRyu16eJKI5J3CogCdO6OCn3/qbdTOqeIvHlrL7T9aTXNHOt9licg4prAoUNUTUvzHxy7hjivOYvma3Vz7rWd5cXtjvssSkXFKYVHA4rGg4/uBT1xKe1c313/nt3zl/62nrbM736WJyDijsBgFLp03mUduX8KNi2fzvWe2cs0/Ps0LWw/muywRGUcUFqPEhOIivvqBC3jgTy4h3eN86Lu/43M/WcM+XWIrIiNAYTHKvPXMKTzy2SX86TvmsXz1bt75d0/ynSc305HWqSkRiY7CYhQqSyX4/DXn8ujtS3jrmVP4+sMbuPIfnuYXq3fRoyHPRSQCCotRbM6UMr73R7V8/+OLKSmK85kHV3P1Pz7Nw6+8oXszRGRYKSzGgMvmV7Pi05fxrZsuorvHufU/X+R933qWFWvf0MOVRGRYRBoWZna1mb1mZvVmdmcfy5eY2Ytmljaz67OW3Wxmm8LXzVHWORbEYsb7LzyNR29/B/d86EJaOtL82Q9e5F1//yTf/902XW4rIqfEojpdYWZxYCNwBbATWAnc5O7rM9aZA1QAnwOWu/tPw/ZJQB1QCziwCnizu/d7V1ptba3X1dVFsi+jUXeP8+i6PXz36S2s3nGIqtIiblg0mw8vns3syaX5Lk9ECoSZrXL32lzrRTlE+WKg3t23hAU9CCwFjoaFu28Ll/VkbXsV8Ji7HwyXPwZcDfwwwnrHlHjMuOaCGVx9/nRWbmvkX57ZwrKnN3PfU5tZclY1f3jJbN59zlQ9O0NEBiXKsKgBdmTM7wQuOYVta7JXMrNbgFsAZs+efXJVjnFmxuK5k1g8dxJvHG7jRyt38OALO/jT769iWkWKD148kw9cVMNZ0ybku1QRKWCj+uFH7r4MWAbBaag8l1PwZkws4bOXn8Vt7zyTxzfs44cvbGfZ01v4zpObOXdGBR+46DSuvbCG6ROL812qiBSYKMNiFzArY35m2DbYbX8va9snh6UqIRGPceV507nyvOnsb+7gl2t287PVu/nqig38za82sOj0SVx1/nSuOm8aM6vUvyEi0XZwJwg6uN9N8Mt/JfBhd1/Xx7r/Bvwyq4N7FXBxuMqLBB3c/Q6IpA7uU7d1fwu/WL2Lh1/Zw4Y9RwA4v6aCqxZM5/IF0zhn+gTMLM9VishwGmwHd2RhERbxHuCbQBy4392/YmZ3A3XuvtzMFgE/A6qAdmCPu58Xbvsx4AvhR33F3f/PQD9LYTG8tu1v4ZF1e3hk3R5e3H4IgGkVKZbMr+YdZ1fz9jOnUFmazHOVInKqCiIsRpLCIjr7mtp5cmMDT21s4JmNDTS1p4kZLJxVydvnV3Pp3ElcfHoVxUXxfJcqIkOksJBIpLt7WLPzME+F4bF25yF6HIrixoUzK7lk3iQumTuZN59eRVlqVF8/ITIuKCxkRDS1d7FqWyPPbT3A81sOsnbXYbp7nETMOK9mIhfNquTCWRNZOKuKOZNL1echUmAUFpIXLR1pVr3eyPNbD1C3rZG1uw7TGg41UllaxIUzK1k4K3idd1oF1RNSChCRPCqEO7hlHCpLJVhyVjVLzqoGgtNWm/Y1s3rHIdbsOMTqHYf41uOb6B3fcEp5knNnVLBgRgULTgve504p053lIgVGRxYy4po70qzbdZhX32hiffjauKeZzu5g1JdUIsbZ0ycwf+oEzpxazvyp5cyfVs7MqlLiMR2FiAwnHVlIwSpPJbhk3mQumTf5aFtXdw+bG5pZv7uJ9bubeHVPE8/WN/DQizuPrpNMxDijuvxogJw5tZy5U8o4fXIppUn9VRaJkv6FSUEoisc4Z3oF50yv4IMXH2tvau+ifl8z9XubqW9oZtPeI6ze0cgvX95N5kHx1Akp5kwOgmPOlDJmTyoN5qeUUlFcNPI7JDLGKCykoFUUF3Hx7Counl11XHtbZzebG5p5/UAr2w608PqBFrYdaOXpTQ38ZNXO49adWFJETWUJNVUl1FSWMPPoeyk1VSVUlRapk10kB4WFjEolyTjn10zk/JqJJyxr7Uyz/WAr2/YHQbKzsZVdjW28fqCF39bvpyXrQVAlRfGjQXJaZTHTKoLX9Ire6RSTypIKFBnXFBYy5pQmE0dPaWVzdw63dbGzsY1dh9rYlfW+bncTB1o6yL7uIxmPUT0hxfSJQXhMywiSyWUpppSnmFKeZFJZUldyyZiksJBxxcyoLE1SWZrs86gEgs72hiMd7GlqZ+/hdvY2tbOnqYN9Te3saWpnw54jPL1xP80d6T63ryotYnIYHpPLU1SXp5hcljyubUp5UENFcUJHLDIqKCxEshTFY5xWWcJplSUDrtfckWZfUzsHWjo50NxBQ3Pwvr+5gwPNnexv7uDV3U083dzBkfa+gyUeMypLiphYWkRVaZKq0iImlgTvVWVJKkuLqAznK0uTVJUF8yVJjcMlI0thIXKSylMJyqvLmVede92OdDcHWzrZf6ST/S1BmBxq7aSxtZNDrV0cau2isbWTXYfaWb+7icbWLtq6uvv9vFQiRlVpkoqSBBXFRUwoTlBREr4XFzGhj7aKjPmSoriOaGRIFBYiIyCViDNjYgkzJg58tJKpvaubw21BiDS2dHGotZND4fyh1i4aWzo50p6mqb2L/c2dbNnfEsy3dZHuGfhm20TMjguT8lTwKk0mKEvFKUsmKE0lKE/Fj2srSyUoTcaDdVMJypMJSlNxitRPM+YpLEQKVHFRnOKiONMqhvaYW3envauHpvYujrR3cbgtzZH2Lpraw/ej811Hw6W5I83uQ+20dqZp7uimtTN9dEyvwUjGY5RlBksqEYZLb/AcaytNxilJxikpCl7F4XRxUWZb7GibgqgwKCxExhgzC34ZJ4ceNJl6epzWrm5aO9I0dwTh0dKRpqUzTUsYKM0d4fLONK0d3eGyY+s2HOmgJQye5o40nemeIdeRiNnRUCkuih0LlKJgH4sT4Xtv2CRjFCeC+VRRjFQiRioRJ5WIkeydzmgP2sJXUbBeImY6TZdFYSEifYrF7OjpqanD9Jld3T20dnTTnu6mrbObtq7g1R6+2jp7jrZ1dGWv0xOuc6ytsaXz6HRbZ7i8q5vuHKfhcu67cSxYErEwXDIDJ3PZ8e3JeDBdFA9eyUSMZNyOmw/eg7ZkPEbRCdvZCZ9TFM9vgCksRGTEFMVjTCyNMZFoh2DpTPfQke6mI90TTofzXcem+2rP3K4j3UNHVx+fke6ho6uHQ62dfX5+Z3fwimKM1mQYGr3h0hs+551WwT9/+OLcH3AKFBYiMuYkw//pT8hjDenuHrq6nc7uHrq6g1DpOjp9rL0r3UNH+N7V7UfXzd6uM2NZ9ufMnjT4CydOlsJCRCQCiXiMRBxKGBv3xOgyAxERyUlhISIiOSksREQkJ4WFiIjkFGlYmNnVZvaamdWb2Z19LE+Z2Y/C5c+b2ZywfY6ZtZnZ6vB1X5R1iojIwCK7GsrM4sC9wBXATmClmS139/UZq30caHT3M83sRuDrwA3hss3uvjCq+kREZPCiPLJYDNS7+xZ37wQeBJZmrbMU+Pdw+qfAu0332IuIFJwow6IG2JExvzNs63Mdd08Dh4HJ4bK5ZvaSmT1lZpf19QPM7BYzqzOzuoaGhuGtXkREjirUm/LeAGa7+wEzezPwczM7z92bMldy92XAMgAzazCz10/hZ04B9p/C9oVkrOzLWNkP0L4UKu0LnD6YlaIMi13ArIz5mWFbX+vsNLMEMBE44O4OdAC4+yoz2wycBdT198PcfRCPoOmfmdW5e+2pfEahGCv7Mlb2A7QvhUr7MnhRnoZaCcw3s7lmlgRuBJZnrbMcuDmcvh543N3dzKrDDnLMbB4wH9gSYa0iIjKAyI4s3D1tZrcBjwBx4H53X2dmdwN17r4c+Ffg+2ZWDxwkCBSAJcDdZtYF9AC3uvvBqGoVEZGBRdpn4e4rgBVZbXdlTLcDf9DHdg8BD0VZWx+WjfDPi9JY2Zexsh+gfSlU2pdBMo9i0HURERlTNNyHiIjkNO7DIteQJIXOzLaZ2dpwWJS6sG2SmT1mZpvC96p819kXM7vfzPaZ2SsZbX3WboF/Cr+nl80s2seCDVE/+/JXZrYrY9ia92Qs+3y4L6+Z2VX5qbpvZjbLzJ4ws/Vmts7MPhO2j6rvZoD9GHXfi5kVm9kLZrYm3Jcvh+1zw6GS6sOhk5Jhe59DKZ0Sdx+3L4KO983APCAJrAEW5LuuIe7DNmBKVts3gDvD6TuBr+e7zn5qXwJcDLySq3bgPcCvAAMuBZ7Pd/2D2Je/Aj7Xx7oLwr9rKWBu+Hcwnu99yKhvBnBxOD0B2BjWPKq+mwH2Y9R9L+GfbXk4XQQ8H/5Z/xi4MWy/D/hkOP1nwH3h9I3Aj061hvF+ZDGYIUlGo8xhVP4duC6PtfTL3Z8muAouU3+1LwX+wwPPAZVmNmNkKs2tn33pz1LgQXfvcPetQD3B38WC4O5vuPuL4fQR4FWC0RZG1XczwH70p2C/l/DPtjmcLQpfDryLYKgkOPE7GdahlMZ7WAxmSJIsXcqBAAAEe0lEQVRC58CjZrbKzG4J26a5+xvh9B5gWn5KOyn91T5av6vbwlMz92ecDhw1+xKevriI4H+yo/a7ydoPGIXfi5nFzWw1sA94jODI55AHQyXB8fUONJTSSRnvYTEWvN3dLwauAT5lZksyF3pwHDoqL3kbzbWHvgOcASwkGMLm7/NbztCYWTnBJeyf9ROH2hk1300f+zEqvxd37/ZgJO6ZBEc854zkzx/vYTGYIUkKmrvvCt/3AT8j+Eu0t/c0QPi+L38VDll/tY+678rd94b/wHuA73HslEbB74uZFRH8gv2Bu/9X2Dzqvpu+9mM0fy8A7n4IeAJ4C8Epv9775TLrPbovljGU0qn83PEeFoMZkqRgmVmZmU3onQauBF7h+GFUbgZ+kZ8KT0p/tS8H/ii88uZS4HDGKZGClHXe/gME3w0E+3JjeMXKXILhbF4Y6fr6E57b/lfgVXe/J2PRqPpu+tuP0fi9WDAEUmU4XULwnKBXCULj+nC17O/khKGUTqmIfPfy5/tFcCXHRoLzf1/Mdz1DrH0ewdUba4B1vfUTnJv8NbAJ+G9gUr5r7af+HxKcBugiON/68f5qJ7ga5N7we1oL1Oa7/kHsy/fDWl8O//HOyFj/i+G+vAZck+/6s/bl7QSnmF4GVoev94y272aA/Rh13wvwJuClsOZXgLvC9nkEgVYP/ARIhe3F4Xx9uHzeqdagO7hFRCSn8X4aSkREBkFhISIiOSksREQkJ4WFiIjkpLAQEZGcFBYiITNrDt/nmNmHh/mzv5A1/9vh/HyRqCksRE40BxhSWGTcRduf48LC3d86xJpE8kphIXKirwGXhc86uD0cwO1vzWxlOPjcnwKY2e+Z2TNmthxYH7b9PBzUcV3vwI5m9jWgJPy8H4RtvUcxFn72KxY8l+SGjM9+0sx+amYbzOwHvaOGmtnXwmc0vGxmfzfifzoyLkX6DG6RUepOgucdvA8g/KV/2N0XmVkK+I2ZPRquezFwvgdDWgN8zN0PhkMyrDSzh9z9TjO7zYNB4LJ9kGBAuwuBKeE2T4fLLgLOA3YDvwHeZmavEgxRcY67e+8QECJR05GFSG5XEox9tJpgiOvJBOMGAbyQERQAnzazNcBzBAO5zWdgbwd+6MHAdnuBp4BFGZ+904MB71YTnB47DLQD/2pmHwRaT3nvRAZBYSGSmwH/090Xhq+57t57ZNFydCWz3wMuB97i7hcSjOVTfAo/tyNjuhtIePBsgsUED7R5H/DwKXy+yKApLEROdITgMZy9HgE+GQ53jZmdFY7ym20i0OjurWZ2DsFjL3t19W6f5RnghrBfpJrg8az9jnQaPpthoruvAG4nOH0lEjn1WYic6GWgOzyd9G/APxKcAnox7GRuoO9H1T4M3Br2K7xGcCqq1zLgZTN70d3/MKP9ZwTPJVhDMELq/3b3PWHY9GUC8AszKyY44rnj5HZRZGg06qyIiOSk01AiIpKTwkJERHJSWIiISE4KCxERyUlhISIiOSksREQkJ4WFiIjkpLAQEZGc/j+bAK/ynmj6MAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112b0c160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.data import iris_data\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "from mlxtend.classifier import SoftmaxRegression\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Loading Data\n",
    "\n",
    "X, y = iris_data()\n",
    "X = X[:, [0, 3]] # sepal length and petal width\n",
    "\n",
    "# standardize\n",
    "X[:,0] = (X[:,0] - X[:,0].mean()) / X[:,0].std()\n",
    "X[:,1] = (X[:,1] - X[:,1].mean()) / X[:,1].std()\n",
    "\n",
    "lr = SoftmaxRegression(eta=0.01, epochs=300, minibatches=len(y), random_seed=1)\n",
    "lr.fit(X, y)\n",
    "\n",
    "plot_decision_regions(X, y, clf=lr)\n",
    "plt.title('Softmax Regression - Stochastic Gradient Descent')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(lr.cost_)), lr.cost_)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# StackingClassifier\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/StackingClassifier_files/stackingclassification_overview.png)\n",
    "\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/StackingClassifier_files/stacking_algorithm.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:56:14.223825Z",
     "start_time": "2018-04-19T10:56:14.092995Z"
    }
   },
   "source": [
    "## Example 1 - Simple Stacked Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:57:06.789258Z",
     "start_time": "2018-04-19T10:57:06.777736Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, 1:3], iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:57:25.811307Z",
     "start_time": "2018-04-19T10:57:25.680808Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3-fold cross validation:\n",
      "\n",
      "Accuracy: 0.91 (+/- 0.01) [KNN]\n",
      "Accuracy: 0.91 (+/- 0.06) [Random Forest]\n",
      "Accuracy: 0.92 (+/- 0.03) [Naive Bayes]\n",
      "Accuracy: 0.95 (+/- 0.03) [StackingClassifier]\n"
     ]
    }
   ],
   "source": [
    "from sklearn import model_selection\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from mlxtend.classifier import StackingClassifier\n",
    "import numpy as np\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "sclf = StackingClassifier(classifiers=[clf1, clf2, clf3], meta_classifier=lr)\n",
    "\n",
    "print('3-fold cross validation:\\n')\n",
    "\n",
    "for clf, label in zip(\n",
    "    [clf1, clf2, clf3, sclf],\n",
    "    ['KNN', 'Random Forest', 'Naive Bayes', 'StackingClassifier']):\n",
    "\n",
    "    scores = model_selection.cross_val_score(clf, X, y, cv=3, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:57:42.241586Z",
     "start_time": "2018-04-19T10:57:39.913389Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112bf5748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "import matplotlib.gridspec as gridspec\n",
    "import itertools\n",
    "\n",
    "gs = gridspec.GridSpec(2, 2)\n",
    "\n",
    "fig = plt.figure(figsize=(10,8))\n",
    "\n",
    "for clf, lab, grd in zip([clf1, clf2, clf3, sclf], \n",
    "                         ['KNN', \n",
    "                          'Random Forest', \n",
    "                          'Naive Bayes',\n",
    "                          'StackingClassifier'],\n",
    "                          itertools.product([0, 1], repeat=2)):\n",
    "\n",
    "    clf.fit(X, y)\n",
    "    ax = plt.subplot(gs[grd[0], grd[1]])\n",
    "    fig = plot_decision_regions(X=X, y=y, clf=clf)\n",
    "    plt.title(lab)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Using Probabilities as Meta-Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:59:17.011422Z",
     "start_time": "2018-04-19T10:59:16.885257Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3-fold cross validation:\n",
      "\n",
      "Accuracy: 0.91 (+/- 0.01) [KNN]\n",
      "Accuracy: 0.91 (+/- 0.06) [Random Forest]\n",
      "Accuracy: 0.92 (+/- 0.03) [Naive Bayes]\n",
      "Accuracy: 0.94 (+/- 0.03) [StackingClassifier]\n"
     ]
    }
   ],
   "source": [
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "sclf = StackingClassifier(\n",
    "    classifiers=[clf1, clf2, clf3],\n",
    "    use_probas=True,\n",
    "    average_probas=False,\n",
    "    meta_classifier=lr)\n",
    "\n",
    "print('3-fold cross validation:\\n')\n",
    "\n",
    "for clf, label in zip(\n",
    "    [clf1, clf2, clf3, sclf],\n",
    "    ['KNN', 'Random Forest', 'Naive Bayes', 'StackingClassifier']):\n",
    "\n",
    "    scores = model_selection.cross_val_score(clf, X, y, cv=3, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 3 - Stacked Classification and GridSearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T10:59:44.610142Z",
     "start_time": "2018-04-19T10:59:42.639514Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.927 +/- 0.02 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.913 +/- 0.03 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.933 +/- 0.02 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.940 +/- 0.02 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Best parameters: {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Accuracy: 0.94\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from mlxtend.classifier import StackingClassifier\n",
    "\n",
    "# Initializing models\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "sclf = StackingClassifier(classifiers=[clf1, clf2, clf3], \n",
    "                          meta_classifier=lr)\n",
    "\n",
    "params = {'kneighborsclassifier__n_neighbors': [1, 5],\n",
    "          'randomforestclassifier__n_estimators': [10, 50],\n",
    "          'meta-logisticregression__C': [0.1, 10.0]}\n",
    "\n",
    "grid = GridSearchCV(estimator=sclf, \n",
    "                    param_grid=params, \n",
    "                    cv=5,\n",
    "                    refit=True)\n",
    "grid.fit(X, y)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\"\n",
    "          % (grid.cv_results_[cv_keys[0]][r],\n",
    "             grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "             grid.cv_results_[cv_keys[2]][r]))\n",
    "\n",
    "print('Best parameters: %s' % grid.best_params_)\n",
    "print('Accuracy: %.2f' % grid.best_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:00:01.746982Z",
     "start_time": "2018-04-19T10:59:57.966354Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.907 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.913 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.927 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.913 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.927 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.913 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.933 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.940 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Best parameters: {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Accuracy: 0.94\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "# Initializing models\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "sclf = StackingClassifier(classifiers=[clf1, clf1, clf2, clf3], \n",
    "                          meta_classifier=lr)\n",
    "\n",
    "params = {'kneighborsclassifier-1__n_neighbors': [1, 5],\n",
    "          'kneighborsclassifier-2__n_neighbors': [1, 5],\n",
    "          'randomforestclassifier__n_estimators': [10, 50],\n",
    "          'meta-logisticregression__C': [0.1, 10.0]}\n",
    "\n",
    "grid = GridSearchCV(estimator=sclf, \n",
    "                    param_grid=params, \n",
    "                    cv=5,\n",
    "                    refit=True)\n",
    "grid.fit(X, y)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\"\n",
    "          % (grid.cv_results_[cv_keys[0]][r],\n",
    "             grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "             grid.cv_results_[cv_keys[2]][r]))\n",
    "\n",
    "print('Best parameters: %s' % grid.best_params_)\n",
    "print('Accuracy: %.2f' % grid.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 4 - Stacking of Classifiers that Operate on Different Feature Subsets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:00:44.843306Z",
     "start_time": "2018-04-19T11:00:44.813010Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StackingClassifier(average_probas=False,\n",
       "          classifiers=[Pipeline(memory=None,\n",
       "     steps=[('columnselector', ColumnSelector(cols=(0, 2), drop_axis=False)), ('logisticregression', LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          pen...='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False))])],\n",
       "          meta_classifier=LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False),\n",
       "          refit=True, store_train_meta_features=False,\n",
       "          use_features_in_secondary=False, use_probas=False, verbose=0)"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.classifier import StackingClassifier\n",
    "from mlxtend.feature_selection import ColumnSelector\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "iris = load_iris()\n",
    "X = iris.data\n",
    "y = iris.target\n",
    "\n",
    "pipe1 = make_pipeline(ColumnSelector(cols=(0, 2)),\n",
    "                      LogisticRegression())\n",
    "pipe2 = make_pipeline(ColumnSelector(cols=(1, 2, 3)),\n",
    "                      LogisticRegression())\n",
    "\n",
    "sclf = StackingClassifier(classifiers=[pipe1, pipe2], \n",
    "                          meta_classifier=LogisticRegression())\n",
    "\n",
    "sclf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# StackingCVClassifier\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/StackingCVClassifier_files/stacking_cv_classification_overview.png)\n",
    "\n",
    "![](http://rasbt.github.io/mlxtend/user_guide/classifier/StackingCVClassifier_files/stacking_cv_algorithm.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 1 - Simple Stacking CV Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:02:10.319129Z",
     "start_time": "2018-04-19T11:02:10.312388Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn import datasets\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X, y = iris.data[:, 1:3], iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:02:21.007577Z",
     "start_time": "2018-04-19T11:02:20.789154Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3-fold cross validation:\n",
      "\n",
      "Accuracy: 0.91 (+/- 0.01) [KNN]\n",
      "Accuracy: 0.90 (+/- 0.03) [Random Forest]\n",
      "Accuracy: 0.92 (+/- 0.03) [Naive Bayes]\n",
      "Accuracy: 0.93 (+/- 0.02) [StackingClassifier]\n"
     ]
    }
   ],
   "source": [
    "from sklearn import model_selection\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from mlxtend.classifier import StackingCVClassifier\n",
    "import numpy as np\n",
    "\n",
    "RANDOM_SEED = 42\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=RANDOM_SEED)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "\n",
    "# The StackingCVClassifier uses scikit-learn's check_cv\n",
    "# internally, which doesn't support a random seed. Thus\n",
    "# NumPy's random seed need to be specified explicitely for\n",
    "# deterministic behavior\n",
    "np.random.seed(RANDOM_SEED)\n",
    "sclf = StackingCVClassifier(classifiers=[clf1, clf2, clf3], \n",
    "                            meta_classifier=lr)\n",
    "\n",
    "print('3-fold cross validation:\\n')\n",
    "\n",
    "for clf, label in zip([clf1, clf2, clf3, sclf], \n",
    "                      ['KNN', \n",
    "                       'Random Forest', \n",
    "                       'Naive Bayes',\n",
    "                       'StackingClassifier']):\n",
    "\n",
    "    scores = model_selection.cross_val_score(clf, X, y, \n",
    "                                              cv=3, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" \n",
    "          % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:02:33.585633Z",
     "start_time": "2018-04-19T11:02:31.129501Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12264d828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from mlxtend.plotting import plot_decision_regions\n",
    "import matplotlib.gridspec as gridspec\n",
    "import itertools\n",
    "\n",
    "gs = gridspec.GridSpec(2, 2)\n",
    "\n",
    "fig = plt.figure(figsize=(10,8))\n",
    "\n",
    "for clf, lab, grd in zip([clf1, clf2, clf3, sclf], \n",
    "                         ['KNN', \n",
    "                          'Random Forest', \n",
    "                          'Naive Bayes',\n",
    "                          'StackingCVClassifier'],\n",
    "                          itertools.product([0, 1], repeat=2)):\n",
    "\n",
    "    clf.fit(X, y)\n",
    "    ax = plt.subplot(gs[grd[0], grd[1]])\n",
    "    fig = plot_decision_regions(X=X, y=y, clf=clf)\n",
    "    plt.title(lab)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 2 - Using Probabilities as Meta-Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:03:11.596029Z",
     "start_time": "2018-04-19T11:03:11.356632Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3-fold cross validation:\n",
      "\n",
      "Accuracy: 0.91 (+/- 0.01) [KNN]\n",
      "Accuracy: 0.91 (+/- 0.06) [Random Forest]\n",
      "Accuracy: 0.92 (+/- 0.03) [Naive Bayes]\n",
      "Accuracy: 0.95 (+/- 0.04) [StackingClassifier]\n"
     ]
    }
   ],
   "source": [
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=1)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "\n",
    "# The StackingCVClassifier uses scikit-learn's check_cv\n",
    "# internally, which doesn't support a random seed. Thus\n",
    "# NumPy's random seed need to be specified explicitely for\n",
    "# deterministic behavior\n",
    "np.random.seed(RANDOM_SEED)\n",
    "sclf = StackingCVClassifier(classifiers=[clf1, clf2, clf3],\n",
    "                            use_probas=True,\n",
    "                            meta_classifier=lr)\n",
    "\n",
    "print('3-fold cross validation:\\n')\n",
    "\n",
    "for clf, label in zip([clf1, clf2, clf3, sclf], \n",
    "                      ['KNN', \n",
    "                       'Random Forest', \n",
    "                       'Naive Bayes',\n",
    "                       'StackingClassifier']):\n",
    "\n",
    "    scores = model_selection.cross_val_score(clf, X, y, \n",
    "                                              cv=3, scoring='accuracy')\n",
    "    print(\"Accuracy: %0.2f (+/- %0.2f) [%s]\" \n",
    "          % (scores.mean(), scores.std(), label))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:03:30.653793Z",
     "start_time": "2018-04-19T11:03:30.650370Z"
    }
   },
   "source": [
    "## Example 3 - Stacked CV Classification and GridSearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:03:50.567168Z",
     "start_time": "2018-04-19T11:03:45.741386Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.673 +/- 0.01 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.920 +/- 0.02 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.893 +/- 0.02 {'kneighborsclassifier__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.947 +/- 0.02 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.947 +/- 0.02 {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Best parameters: {'kneighborsclassifier__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "Accuracy: 0.95\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from mlxtend.classifier import StackingCVClassifier\n",
    "\n",
    "# Initializing models\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=RANDOM_SEED)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "\n",
    "# The StackingCVClassifier uses scikit-learn's check_cv\n",
    "# internally, which doesn't support a random seed. Thus\n",
    "# NumPy's random seed need to be specified explicitely for\n",
    "# deterministic behavior\n",
    "np.random.seed(RANDOM_SEED)\n",
    "sclf = StackingCVClassifier(classifiers=[clf1, clf2, clf3], \n",
    "                            meta_classifier=lr)\n",
    "\n",
    "params = {'kneighborsclassifier__n_neighbors': [1, 5],\n",
    "          'randomforestclassifier__n_estimators': [10, 50],\n",
    "          'meta-logisticregression__C': [0.1, 10.0]}\n",
    "\n",
    "grid = GridSearchCV(estimator=sclf, \n",
    "                    param_grid=params, \n",
    "                    cv=5,\n",
    "                    refit=True)\n",
    "grid.fit(X, y)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\"\n",
    "          % (grid.cv_results_[cv_keys[0]][r],\n",
    "             grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "             grid.cv_results_[cv_keys[2]][r]))\n",
    "\n",
    "print('Best parameters: %s' % grid.best_params_)\n",
    "print('Accuracy: %.2f' % grid.best_score_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:04:12.574026Z",
     "start_time": "2018-04-19T11:04:03.412122Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.673 +/- 0.01 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.920 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.893 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.947 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.940 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 1, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.953 +/- 0.02 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.927 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 10}\n",
      "0.667 +/- 0.00 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 0.1, 'randomforestclassifier__n_estimators': 50}\n",
      "0.940 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "0.940 +/- 0.03 {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 5, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 50}\n",
      "Best parameters: {'kneighborsclassifier-1__n_neighbors': 5, 'kneighborsclassifier-2__n_neighbors': 1, 'meta-logisticregression__C': 10.0, 'randomforestclassifier__n_estimators': 10}\n",
      "Accuracy: 0.95\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "# Initializing models\n",
    "\n",
    "clf1 = KNeighborsClassifier(n_neighbors=1)\n",
    "clf2 = RandomForestClassifier(random_state=RANDOM_SEED)\n",
    "clf3 = GaussianNB()\n",
    "lr = LogisticRegression()\n",
    "\n",
    "# The StackingCVClassifier uses scikit-learn's check_cv\n",
    "# internally, which doesn't support a random seed. Thus\n",
    "# NumPy's random seed need to be specified explicitely for\n",
    "# deterministic behavior\n",
    "np.random.seed(RANDOM_SEED)\n",
    "sclf = StackingCVClassifier(classifiers=[clf1, clf1, clf2, clf3], \n",
    "                            meta_classifier=lr)\n",
    "\n",
    "params = {'kneighborsclassifier-1__n_neighbors': [1, 5],\n",
    "          'kneighborsclassifier-2__n_neighbors': [1, 5],\n",
    "          'randomforestclassifier__n_estimators': [10, 50],\n",
    "          'meta-logisticregression__C': [0.1, 10.0]}\n",
    "\n",
    "grid = GridSearchCV(estimator=sclf, \n",
    "                    param_grid=params, \n",
    "                    cv=5,\n",
    "                    refit=True)\n",
    "grid.fit(X, y)\n",
    "\n",
    "cv_keys = ('mean_test_score', 'std_test_score', 'params')\n",
    "\n",
    "for r, _ in enumerate(grid.cv_results_['mean_test_score']):\n",
    "    print(\"%0.3f +/- %0.2f %r\"\n",
    "          % (grid.cv_results_[cv_keys[0]][r],\n",
    "             grid.cv_results_[cv_keys[1]][r] / 2.0,\n",
    "             grid.cv_results_[cv_keys[2]][r]))\n",
    "\n",
    "print('Best parameters: %s' % grid.best_params_)\n",
    "print('Accuracy: %.2f' % grid.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example 4 - Stacking of Classifiers that Operate on Different Feature Subsets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:05:01.234523Z",
     "start_time": "2018-04-19T11:05:01.207516Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StackingCVClassifier(classifiers=[Pipeline(memory=None,\n",
       "     steps=[('columnselector', ColumnSelector(cols=(0, 2), drop_axis=False)), ('logisticregression', LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          pen...='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False))])],\n",
       "           cv=2,\n",
       "           meta_classifier=LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,\n",
       "          intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,\n",
       "          penalty='l2', random_state=None, solver='liblinear', tol=0.0001,\n",
       "          verbose=0, warm_start=False),\n",
       "           refit=True, shuffle=True, store_train_meta_features=False,\n",
       "           stratify=True, use_features_in_secondary=False,\n",
       "           use_probas=False, verbose=0)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import load_iris\n",
    "from mlxtend.classifier import StackingCVClassifier\n",
    "from mlxtend.feature_selection import ColumnSelector\n",
    "from sklearn.pipeline import make_pipeline\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "iris = load_iris()\n",
    "X = iris.data\n",
    "y = iris.target\n",
    "\n",
    "pipe1 = make_pipeline(ColumnSelector(cols=(0, 2)),\n",
    "                      LogisticRegression())\n",
    "pipe2 = make_pipeline(ColumnSelector(cols=(1, 2, 3)),\n",
    "                      LogisticRegression())\n",
    "\n",
    "sclf = StackingCVClassifier(classifiers=[pipe1, pipe2], \n",
    "                            meta_classifier=LogisticRegression())\n",
    "\n",
    "sclf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Kmeans"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:06:11.162894Z",
     "start_time": "2018-04-19T11:06:11.011643Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1134060f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "from mlxtend.data import three_blobs_data\n",
    "\n",
    "X, y = three_blobs_data()\n",
    "plt.scatter(X[:, 0], X[:, 1], c='b')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:06:16.148822Z",
     "start_time": "2018-04-19T11:06:16.126416Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Iteration: 2/50 | Elapsed: 00:00:00 | ETA: 00:00:00"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iterations until convergence: 2\n",
      "Final centroids:\n",
      " [[-1.5947298   2.92236966]\n",
      " [ 2.06521743  0.96137409]\n",
      " [ 0.9329651   4.35420713]]\n"
     ]
    }
   ],
   "source": [
    "from mlxtend.cluster import Kmeans\n",
    "\n",
    "km = Kmeans(k=3, \n",
    "            max_iter=50, \n",
    "            random_seed=1, \n",
    "            print_progress=3)\n",
    "\n",
    "km.fit(X)\n",
    "\n",
    "print('Iterations until convergence:', km.iterations_)\n",
    "print('Final centroids:\\n', km.centroids_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:06:30.524008Z",
     "start_time": "2018-04-19T11:06:30.291709Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112d8cf98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_clust = km.predict(X)\n",
    "\n",
    "plt.scatter(X[y_clust == 0, 0],\n",
    "            X[y_clust == 0, 1],\n",
    "            s=50,\n",
    "            c='lightgreen',\n",
    "            marker='s',\n",
    "            label='cluster 1')\n",
    "\n",
    "plt.scatter(X[y_clust == 1,0],\n",
    "            X[y_clust == 1,1],\n",
    "            s=50,\n",
    "            c='orange',\n",
    "            marker='o',\n",
    "            label='cluster 2')\n",
    "\n",
    "plt.scatter(X[y_clust == 2,0],\n",
    "            X[y_clust == 2,1],\n",
    "            s=50,\n",
    "            c='lightblue',\n",
    "            marker='v',\n",
    "            label='cluster 3')\n",
    "\n",
    "\n",
    "plt.scatter(km.centroids_[:,0],\n",
    "            km.centroids_[:,1],\n",
    "            s=250,\n",
    "            marker='*',\n",
    "            c='red',\n",
    "            label='centroids')\n",
    "\n",
    "plt.legend(loc='lower left',\n",
    "           scatterpoints=1)\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plotting Learning Curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-04-19T11:11:38.130505Z",
     "start_time": "2018-04-19T11:11:37.825371Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x109c0cbe0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from mlxtend.plotting import plot_learning_curves\n",
    "import matplotlib.pyplot as plt\n",
    "from mlxtend.data import iris_data\n",
    "from mlxtend.preprocessing import shuffle_arrays_unison\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "# Loading some example data\n",
    "X, y = iris_data()\n",
    "X, y = shuffle_arrays_unison(arrays=[X, y], random_seed=123)\n",
    "X_train, X_test = X[:100], X[100:]\n",
    "y_train, y_test = y[:100], y[100:]\n",
    "\n",
    "clf = KNeighborsClassifier(n_neighbors=5)\n",
    "\n",
    "plot_learning_curves(X_train, y_train, X_test, y_test, clf)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "stats36",
   "language": "python",
   "name": "stats36"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}