src/Fashion-MNIST_database_of_fashion_articles_keras_dataset.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Imports\n",
    "\n",
    "`numpy`      import to manupulate arrays  \n",
    "`pandas`     import to create and modify dataframes  \n",
    "`matplotlib` to visulaize graphs  \n",
    "`seaborn`    build on matploblib, higher level graph functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Imports \n",
    "import numpy as np \n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loading the data\n",
    "Here we are taking the inbuilt function of keras to load the data from the server  \n",
    "The dataset file in present in the [Link to dataset in amazon server](http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/)  \n",
    "The inbuilt code \n",
    "```python\n",
    "def load_data():\n",
    "    \"\"\"Loads the Fashion-MNIST dataset.\n",
    "    # Returns\n",
    "        Tuple of Numpy arrays: `(x_train, y_train), (x_test, y_test)`.\n",
    "    \"\"\"\n",
    "    dirname = os.path.join('datasets', 'fashion-mnist')\n",
    "    base = 'http://fashion-mnist.s3-website.eu-central-1.amazonaws.com/'\n",
    "    files = ['train-labels-idx1-ubyte.gz', 'train-images-idx3-ubyte.gz',\n",
    "             't10k-labels-idx1-ubyte.gz', 't10k-images-idx3-ubyte.gz']\n",
    "\n",
    "    paths = []\n",
    "    for fname in files:\n",
    "        paths.append(get_file(fname,\n",
    "                              origin=base + fname,\n",
    "                              cache_subdir=dirname))\n",
    "\n",
    "    with gzip.open(paths[0], 'rb') as lbpath:\n",
    "        y_train = np.frombuffer(lbpath.read(), np.uint8, offset=8)\n",
    "\n",
    "    with gzip.open(paths[1], 'rb') as imgpath:\n",
    "        x_train = np.frombuffer(imgpath.read(), np.uint8,\n",
    "                                offset=16).reshape(len(y_train), 28, 28)\n",
    "\n",
    "    with gzip.open(paths[2], 'rb') as lbpath:\n",
    "        y_test = np.frombuffer(lbpath.read(), np.uint8, offset=8)\n",
    "\n",
    "    with gzip.open(paths[3], 'rb') as imgpath:\n",
    "        x_test = np.frombuffer(imgpath.read(), np.uint8,\n",
    "                               offset=16).reshape(len(y_test), 28, 28)\n",
    "\n",
    "    return (x_train, y_train), (x_test, y_test)\n",
    "```\n",
    "which downloads the data and unpacks and gives back as tuples of test, train splits   \n",
    "\n",
    "## Regarding Dataset\n",
    "Fashion Mnist is data set of 10 fashoin articles namely  \n",
    "\n",
    "| Label | Description   |\n",
    "|:----  |:------------- | \n",
    "|0      | T-shirt/top   | \n",
    "|1      | Trouser       | \n",
    "|2      | Pullover      |\n",
    "|3      | Dress         |\n",
    "|4      | Coat          |\n",
    "|5      | Sandal        |\n",
    "|6      | Shirt         |\n",
    "|7      | Sneaker       |\n",
    "|8      | Bag           |\n",
    "|9      | Ankle boot    |\n",
    "\n",
    "[official Link](https://github.com/zalandoresearch/fashion-mnist)   \n",
    "\n",
    "`Line 4` : created a dictionary with respective labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Loading the dataset\n",
    "from keras.datasets import fashion_mnist\n",
    "(x_train,y_train),(x_test,y_test) = fashion_mnist.load_data()\n",
    "fashion_mnist_labels_dict = {0:\"T-shirt/top\",1:\"Trouser\",2:\"Pullover\",3:\"Dress\",4:\"Coat\",5:\"Sandal\",6:\"Shirt\",7:\"Sneaker\",8:\"Bag\",9:\"Ankle boot\"}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Understand the dataset\n",
    "Let see the size and shape of test and training tuples\n",
    "\n",
    "``On Execution``  \n",
    "There will be 60000 samples of 28*28 resolution of images in training set  \n",
    "There will be 10000 samples of 28*28 resolution of images in testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The number of training samples 60000\n",
      "The number of testing samples 10000\n",
      "The shape of training sample array (60000, 28, 28)\n",
      "The shape of training labels (60000,)\n"
     ]
    }
   ],
   "source": [
    "# Understanding the data\n",
    "print(\"The number of training samples\",len(x_train))\n",
    "print(\"The number of testing samples\",len(x_test))\n",
    "print(\"The shape of training sample array\",np.shape(x_train))\n",
    "print(\"The shape of training labels\",np.shape(y_train))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualizing the data\n",
    "Let us try to visualise the few samples of data so, we could get an idea of how the data looks like  \n",
    "`On Execution`  \n",
    "We can see ten images of "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "first ten labels\n",
      "Label value : 9\n",
      "Object Name : Ankle boot\n",
      "Label value : 0\n",
      "Object Name : T-shirt/top\n",
      "Label value : 0\n",
      "Object Name : T-shirt/top\n",
      "Label value : 3\n",
      "Object Name : Dress\n",
      "Label value : 0\n",
      "Object Name : T-shirt/top\n",
      "Label value : 2\n",
      "Object Name : Pullover\n",
      "Label value : 7\n",
      "Object Name : Sneaker\n",
      "Label value : 2\n",
      "Object Name : Pullover\n",
      "Label value : 5\n",
      "Object Name : Sandal\n",
      "Label value : 5\n",
      "Object Name : Sandal\n"
     ]
    }
   ],
   "source": [
    "# Visulizing the data\n",
    "fig, ax = plt.subplots(2, 5, sharex=True, sharey=True)\n",
    "index = 0\n",
    "for row in ax:\n",
    "    for col in row:\n",
    "        col.set_xlabel(str(fashion_mnist_labels_dict[y_train[index]]))\n",
    "        col.imshow(x_train[index],cmap='gray')\n",
    "        index+=1\n",
    "plt.show()\n",
    "print(\"first ten labels\")\n",
    "for i in range(0,10):\n",
    "    print(\"Label value :\",y_train[i])\n",
    "    print(\"Object Name :\",fashion_mnist_labels_dict[y_train[i]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preprocessing the data\n",
    "\n",
    "## Vairables\n",
    "`image_width`,`image_height`,`image_channels` would describe the dimentions of the image  \n",
    "`classes` to detemine how many catogories of samples present in out dataset. By nature mnist have 0-9 images to ten classes  \n",
    "\n",
    "## Creating sparse vector representation\n",
    "`to_categorical` is converting into one hot encoding. Means each vector is represented by one hot encoding.\n",
    "0 --> [1,0,0,0,0,0,0,0,0,0]   \n",
    "1 --> [0,1,0,0,0,0,0,0,0,0]  \n",
    "2 --> [0,0,1,0,0,0,0,0,0,0]   \n",
    "and similarly goes on \n",
    "\n",
    "`np.expand_dims` would just increst one dimentions in the end. Like .... [1,2,3] to [[1],[2],[3]]  \n",
    "\n",
    "## Normalization\n",
    "`Line16`,`Line17`is normalization we are divinding all the pixal values by 255. so all the numerical values are converted between 0 and 1\n",
    ">Note: since its a simple dataset there is not much of processing required to attain good accuracies. For all real time datasest preprocessing like normalizing , standadising , on hot encoding, filling the missing values, transforming features, feeding the data in batches and all other type of preprocessing is required\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'np' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-1-0f675961ee66>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      7\u001b[0m \u001b[0mimage_shape\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mimage_width\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mimage_height\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mimage_channels\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      8\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 9\u001b[1;33m \u001b[0mx_train\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexpand_dims\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0maxis\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     10\u001b[0m \u001b[0mx_test\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexpand_dims\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx_test\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0maxis\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     11\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'np' is not defined"
     ]
    }
   ],
   "source": [
    "# Preprocessing the data\n",
    "    \n",
    "# Variables \n",
    "image_width  = 28\n",
    "image_height = 28\n",
    "image_channels = 1\n",
    "image_shape = (image_width,image_height,image_channels)\n",
    "\n",
    "x_train = np.expand_dims(x_train,axis=3)\n",
    "x_test = np.expand_dims(x_test,axis=3)\n",
    "\n",
    "## Creating sparse vector representation\n",
    "from keras.utils import to_categorical\n",
    "y_train_sparse = to_categorical(y_train)\n",
    "y_test_sparse  = to_categorical(y_test)\n",
    "\n",
    "## Normalization\n",
    "x_train = x_train /255\n",
    "x_test = x_test/255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training varibles\n",
    "These Training varbles are hyper parameters for neural network training.   \n",
    "`epochs` : each epoch is forward propagation + backward propagation over the whole dataset once is called one epoch.  \n",
    "`learning_rate` : the magnitude in which the weights are modified one the acquired loss.   \n",
    "`learning_rate_decay` : there can be high leanring rate at the beining of the training when the loss is high. Over a period of time the learning rate can reduce for fine training of network.  \n",
    "`batch_size` : the data is fed to the network in batches of 32 samples at each time. This batch feeding is done all over the whole dataset.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Training varibles\n",
    "learning_rate = 0.001\n",
    "learning_rate_decay = 0.00001\n",
    "batch_size = 32\n",
    "epochs = 20\n",
    "classes = 10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural Netowork Model\n",
    "`Line 6` : we are building a keras sequential model  \n",
    "`Line 32` : we are using stochastic gradient decent optimizer  \n",
    "`Line 36` : compiling the model to check if the model is build properly.  \n",
    "\n",
    "The loss function being used is `categorical_crossentropy` since its a multi class classification\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Building the models\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D,MaxPooling2D,Activation,Dropout,Flatten,Dense\n",
    "from keras.optimizers import SGD\n",
    "\n",
    "model = Sequential()\n",
    "\n",
    "# Layer 1\n",
    "model.add(Conv2D(filters=128, kernel_size=(3,3), strides=(1, 1), padding='valid', data_format=\"channels_last\",\n",
    "                input_shape=image_shape))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "# Layer 2\n",
    "model.add(Conv2D(filters = 64, kernel_size=(3,3), strides=(1, 1), padding='valid'))\n",
    "model.add(Activation('relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "# Layer 3\n",
    "model.add(Flatten(data_format=\"channels_last\"))\n",
    "model.add(Dense(512))\n",
    "model.add(Activation('relu'))\n",
    "model.add(Dropout(rate=0.25))\n",
    "\n",
    "# Layer 4\n",
    "model.add(Dense(10,activation=\"softmax\"))\n",
    "\n",
    "sgd_optimizer = SGD(lr = learning_rate, decay = learning_rate_decay)\n",
    "\n",
    "model.compile(optimizer=sgd_optimizer, loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training \n",
    "Training is the process of feeding the data to neural network and modifiying the weights of the model using the the backpropagation algorithm. The backpropagation using loss the function acquires the loss over batch size of data and does a backpropagation to modify the weights in such a way the in the next epoch the loss would be less when compared to the current epoch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples, validate on 10000 samples\n",
      "Epoch 1/20\n",
      "60000/60000 [==============================] - 15s 246us/step - loss: 2.0516 - acc: 0.2741 - val_loss: 1.4144 - val_acc: 0.5728\n",
      "Epoch 2/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 1.2472 - acc: 0.5253 - val_loss: 0.9791 - val_acc: 0.6371\n",
      "Epoch 3/20\n",
      "60000/60000 [==============================] - 12s 201us/step - loss: 1.0451 - acc: 0.6025 - val_loss: 0.8966 - val_acc: 0.6655\n",
      "Epoch 4/20\n",
      "60000/60000 [==============================] - 12s 200us/step - loss: 0.9589 - acc: 0.6394 - val_loss: 0.8358 - val_acc: 0.6880\n",
      "Epoch 5/20\n",
      "60000/60000 [==============================] - 12s 198us/step - loss: 0.8906 - acc: 0.6677 - val_loss: 0.7934 - val_acc: 0.7079\n",
      "Epoch 6/20\n",
      "60000/60000 [==============================] - 12s 198us/step - loss: 0.8429 - acc: 0.6838 - val_loss: 0.7588 - val_acc: 0.7193\n",
      "Epoch 7/20\n",
      "60000/60000 [==============================] - 12s 196us/step - loss: 0.8070 - acc: 0.6976 - val_loss: 0.7402 - val_acc: 0.7262\n",
      "Epoch 8/20\n",
      "60000/60000 [==============================] - 12s 196us/step - loss: 0.7764 - acc: 0.7086 - val_loss: 0.7124 - val_acc: 0.7347\n",
      "Epoch 9/20\n",
      "60000/60000 [==============================] - 12s 196us/step - loss: 0.7537 - acc: 0.7179 - val_loss: 0.6880 - val_acc: 0.7424\n",
      "Epoch 10/20\n",
      "60000/60000 [==============================] - 12s 196us/step - loss: 0.7303 - acc: 0.7265 - val_loss: 0.6741 - val_acc: 0.7483\n",
      "Epoch 11/20\n",
      "60000/60000 [==============================] - 12s 196us/step - loss: 0.7119 - acc: 0.7324 - val_loss: 0.6585 - val_acc: 0.7530\n",
      "Epoch 12/20\n",
      "60000/60000 [==============================] - 12s 200us/step - loss: 0.6986 - acc: 0.7378 - val_loss: 0.6532 - val_acc: 0.7567\n",
      "Epoch 13/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 0.6832 - acc: 0.7426 - val_loss: 0.6333 - val_acc: 0.7613\n",
      "Epoch 14/20\n",
      "60000/60000 [==============================] - 12s 200us/step - loss: 0.6750 - acc: 0.7458 - val_loss: 0.6244 - val_acc: 0.7686\n",
      "Epoch 15/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 0.6608 - acc: 0.7501 - val_loss: 0.6131 - val_acc: 0.7704\n",
      "Epoch 16/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 0.6514 - acc: 0.7550 - val_loss: 0.6059 - val_acc: 0.7715\n",
      "Epoch 17/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 0.6428 - acc: 0.7585 - val_loss: 0.5968 - val_acc: 0.7789\n",
      "Epoch 18/20\n",
      "60000/60000 [==============================] - 12s 199us/step - loss: 0.6362 - acc: 0.7613 - val_loss: 0.5909 - val_acc: 0.7832\n",
      "Epoch 19/20\n",
      "60000/60000 [==============================] - 12s 200us/step - loss: 0.6255 - acc: 0.7655 - val_loss: 0.5804 - val_acc: 0.7848\n",
      "Epoch 20/20\n",
      "60000/60000 [==============================] - 12s 200us/step - loss: 0.6213 - acc: 0.7654 - val_loss: 0.5760 - val_acc: 0.7872\n"
     ]
    }
   ],
   "source": [
    "# Training the mode\n",
    "model_history = model.fit(x=x_train, y=y_train_sparse, batch_size=batch_size, epochs=epochs, verbose=1, validation_data=(x_test,y_test_sparse), shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Results\n",
    "using the trained model we try to predict what are the values of images in the test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 72us/step\n"
     ]
    }
   ],
   "source": [
    "# Results\n",
    "y_pred = model.predict(x_test, batch_size = batch_size, verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Verifying the results\n",
    "cheking the results how good they are with the first 10 samples.   \n",
    "Plotting the graphs of test and train set accuracies and loss values. \n",
    "> NOTE: This plot is a very curicial step. These plots would tell us how good the model converges and if there is any overfitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Ground truths of first 10 images in test set [9 2 1 1 6 1 4 6 5 7]\n",
      "Predicted values of first 10 image in test set [9 2 1 1 2 1 6 6 5 7]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Verifying the results\n",
    "print(\"Ground truths of first 10 images in test set\",np.array(y_test[0:10]))\n",
    "print(\"Predicted values of first 10 image in test set\",np.argmax(y_pred[0:10],axis=1))\n",
    "\n",
    "loss = model_history.history['loss']\n",
    "val_loss = model_history.history['val_loss']\n",
    "plt.plot(loss,label='train')\n",
    "plt.plot(val_loss,label='test')\n",
    "plt.title('loss Graph')\n",
    "plt.ylabel('precentage')\n",
    "plt.xlabel('epochs')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "acc = model_history.history['acc']\n",
    "val_acc = model_history.history['val_acc']\n",
    "plt.plot(acc,label='train')\n",
    "plt.plot(val_acc,label='test')\n",
    "plt.title('Accuracy Graph')\n",
    "plt.ylabel('precentage')\n",
    "plt.xlabel('epochs')\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visulizing the results\n",
    "checking the results by visulizing them and creating a confusion matrix. The values of precession and accuracy can be obtained by the help of confusion matrix and f1 scores to compare this architecure with other architectures of neural networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             T-shirt/top  Trouser  Pullover  Dress  Coat  Sandal  Shirt  \\\n",
      "T-shirt/top          796        6        44     72    11       7     53   \n",
      "Trouser                2      941         2     38     8       1      6   \n",
      "Pullover              10        3       691     10   206       2     70   \n",
      "Dress                 32       11        35    838    52       0     29   \n",
      "Coat                   1        1       160     40   731       1     59   \n",
      "Sandal                 1        0         0      2     0     913      0   \n",
      "Shirt                226        2       302     50   184       3    215   \n",
      "Sneaker                0        0         0      0     0      38      0   \n",
      "Bag                    5        3        26      7    14       7     20   \n",
      "Ankle boot             0        1         0      0     2      13      0   \n",
      "\n",
      "             Sneaker  Bag  Ankle boot  \n",
      "T-shirt/top        0   11           0  \n",
      "Trouser            0    2           0  \n",
      "Pullover           0    8           0  \n",
      "Dress              0    3           0  \n",
      "Coat               0    7           0  \n",
      "Sandal            57    3          24  \n",
      "Shirt              0   18           0  \n",
      "Sneaker          905    1          56  \n",
      "Bag                4  912           2  \n",
      "Ankle boot        53    1         930  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(159.0, 0.5, 'Actaul')"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visulizing the results\n",
    "y_pred = np.argmax(y_pred,axis=1)\n",
    "y_pred = pd.Series(y_pred,name = \"predicted\")\n",
    "y_test = pd.Series(y_test,name = \"Actual\")\n",
    "df_confusion  = pd.crosstab(y_test,y_pred)\n",
    "df_confusion.columns = [i for i in list(fashion_mnist_labels_dict.values())]\n",
    "df_confusion.index = [i for i in list(fashion_mnist_labels_dict.values())]\n",
    "\n",
    "print(df_confusion)\n",
    "plt.figure(figsize = (20,20))\n",
    "plt.title('Confusion Matrix',fontsize=20)\n",
    "sns.heatmap(df_confusion, annot=True,fmt=\"d\")\n",
    "plt.xlabel('Predicted', fontsize=18)\n",
    "plt.ylabel('Actaul', fontsize=18)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}