NMF_RS.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "\n",
    "from surprise import NMF\n",
    "from surprise import Dataset\n",
    "from surprise import accuracy\n",
    "from surprise import Reader\n",
    "\n",
    "from surprise.model_selection import GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the Python libraries\n",
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from surprise.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "file_path = os.path.expanduser('~/PycharmProjects/Tesis1/ml-100k/u.data')\n",
    "\n",
    "reader = Reader(line_format='user item rating timestamp', sep='\\t', rating_scale=(1, 5))\n",
    "\n",
    "data = Dataset.load_from_file(file_path, reader=reader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "#Asignamos los datos a una lista raw_ratings, sin indices\n",
    "raw_ratings = data.raw_ratings\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "random.shuffle(raw_ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Separamos en train y test\n",
    "threshold = int(.9 * len(raw_ratings))\n",
    "train_raw_ratings = raw_ratings[:threshold]\n",
    "test_raw_ratings = raw_ratings[threshold:]\n",
    "\n",
    "data.raw_ratings = train_raw_ratings  # Reemplaza en data con los valores de entrenamiento"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Grid Search...\n"
     ]
    }
   ],
   "source": [
    "# Utilizamos gridsearch para obetener los mejores parametros para el algoritmo\n",
    "print('Grid Search...')\n",
    "param_grid = {'n_factors': [50,100,150],'n_epochs':[25,50,75]}\n",
    "grid_search = GridSearchCV(NMF, param_grid, measures=['rmse'], cv=3)\n",
    "grid_search.fit(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9774351799518451\n",
      "{'n_factors': 100, 'n_epochs': 75}\n"
     ]
    }
   ],
   "source": [
    "algo = grid_search.best_estimator['rmse']\n",
    "print(grid_search.best_score['rmse'])\n",
    "print(grid_search.best_params['rmse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<surprise.prediction_algorithms.matrix_factorization.NMF object at 0x7fb0f8053fd0>\n"
     ]
    }
   ],
   "source": [
    "print(algo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<surprise.prediction_algorithms.matrix_factorization.NMF at 0x7fb0f8053fd0>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Se reentrena el set de entrenamiento con el mejor conjunto de parametros obtenido\n",
    "trainset = data.build_full_trainset()\n",
    "algo.fit(trainset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on Trainset,   RMSE: 0.7220\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7220002371925545"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "# Con la primer funcion creamos un set de test a partir del set de entrenamiento\n",
    "predictions = algo.test(trainset.build_testset())\n",
    "print('Accuracy on Trainset,', end='   ')\n",
    "accuracy.rmse(predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on Testset, RMSE: 0.9607\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.9607096199028853"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Prueba el modelo con el set de test\n",
    "testset = data.construct_testset(test_raw_ratings)  # testset is now the set B\n",
    "predictions = algo.test(testset)\n",
    "print('Accuracy on Testset,', end=' ')\n",
    "accuracy.rmse(predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "943\n",
      "1668\n",
      "(1668, 100)\n",
      "(943, 100)\n"
     ]
    }
   ],
   "source": [
    "print(trainset.n_users)\n",
    "print(trainset.n_items)\n",
    "print(algo.qi.shape)\n",
    "print(algo.pu.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}