notes/rs/rs_Deep-Neural-Network.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "from tqdm import tqdm\n",
    "from datetime import datetime\n",
    "from heapq import heappush,heappop\n",
    "from collections import Counter,defaultdict\n",
    "import multiprocessing as mp\n",
    "import tensorflow as tf\n",
    "from sklearn.utils import shuffle\n",
    "import time\n",
    "\n",
    "PATH=\"/home/yui/Documents/data/recommender/movieLens20M/rating.csv\"\n",
    "df = pd.read_csv(PATH)\n",
    "\n",
    "def findNMatches(df,targetUser,leastCommon=25,minMoviesWatched=50,noRef=25):\n",
    "    tardf = df[df[\"userId\"]==targetUser]\n",
    "    u2m = defaultdict(dict)\n",
    "    valCountsUserId = df[\"userId\"].value_counts()\n",
    "\n",
    "    for j in range(tardf.shape[0]):\n",
    "        movieId = tardf.iloc[j][\"movieId\"]\n",
    "        u2m[targetUser][movieId] = tardf.iloc[j][\"rating\"]\n",
    "    avg = np.array(list(u2m[targetUser].values())).mean()\n",
    "    u2m[targetUser] = (u2m[targetUser],avg)\n",
    "    tarSet = set(u2m[targetUser][0])\n",
    "    # print(tarSet)\n",
    "    prevId,storeSet,storeDict = 0,set(),dict()\n",
    "    for i in tqdm(range(df.shape[0])):\n",
    "        row = df.iloc[i]\n",
    "        userId = row[\"userId\"]\n",
    "        if valCountsUserId[userId]<minMoviesWatched:\n",
    "            continue\n",
    "        if userId==targetUser:\n",
    "            continue\n",
    "        if userId!=prevId:\n",
    "            if len(tarSet&storeSet)>=leastCommon:\n",
    "                avg = np.array(list(storeDict.values())).mean()\n",
    "                u2m[prevId]=(storeDict,avg)\n",
    "            storeSet,storeDict=set(),dict()\n",
    "        prevId = userId\n",
    "        storeSet.add(row[\"movieId\"])\n",
    "        storeDict[row[\"movieId\"]]=row[\"rating\"]\n",
    "        if len(u2m)>noRef:\n",
    "            break\n",
    "    return u2m\n",
    "\n",
    "class MF:\n",
    "    def __init__(self,u2m,K=100,penalty=0.1):\n",
    "        self.u2m = u2m\n",
    "        self.penalty = penalty\n",
    "        self.K = K\n",
    "        self.allmovies = dict()\n",
    "        self.mean,i=0,0\n",
    "        for candidate in u2m:\n",
    "            for movie in u2m[candidate][0]:\n",
    "                t,c = self.allmovies.get(movie,(0,0))\n",
    "                c+=1\n",
    "                t+=u2m[candidate][0][movie]\n",
    "                self.allmovies[movie]=(t,c)\n",
    "                i+=1\n",
    "                self.mean+=u2m[candidate][0][movie]\n",
    "        self.mean/=i\n",
    "        for movie in self.allmovies:\n",
    "            t,c = self.allmovies[movie]\n",
    "            self.allmovies[movie] = (t/c,c)\n",
    "        self.m2id = {key:i for i,key in enumerate(self.allmovies)}\n",
    "        self.id2m = {i:key for i,key in enumerate(self.allmovies)}\n",
    "        self.u2id = {key:i for i,key in enumerate(u2m)}\n",
    "        self.id2u = {i:key for i,key in enumerate(u2m)}\n",
    "        self.N = len(u2m)\n",
    "        self.M = len(self.allmovies)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Deep Neural Network\n",
    "In this example, the relationship between user vector $w_i$ and movie vector $u_j$ is not assumed, but is learnt from deep neural network model. \n",
    "\n",
    "$$y = r_{ij}-\\mu = f(w_i,u_j)$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class dnnMF:\n",
    "    def __init__(self,df,targetUser,K=100,\n",
    "                 split=0.85,batch_size=16,hidden=128):\n",
    "        self.u2m = findNMatches(df,targetUser)\n",
    "        self.mf = MF(self.u2m,K=K)\n",
    "        tdf = df[df[\"userId\"].isin(list(self.u2m.keys()))]\n",
    "        self.tdf = shuffle(tdf)\n",
    "        self.split = split\n",
    "        self.batch_size = batch_size\n",
    "        self.hidden = hidden\n",
    "        self.prepareData()\n",
    "        self.getModel()\n",
    "    def getX1X2Y(self,dataset):\n",
    "        w = [self.mf.u2id[i] for i in list(dataset['userId'])]\n",
    "        XW = np.array(w).reshape(-1,1,1)\n",
    "        u = [self.mf.m2id[i] for i in list(dataset['movieId'])]\n",
    "        XU = np.array(u).reshape(-1,1,1)\n",
    "        Y = np.array(dataset['rating'])-self.mf.mean\n",
    "        return XW,XU,Y\n",
    "    def prepareData(self):\n",
    "        trainSet = self.tdf.iloc[:int(self.split*self.tdf.shape[0])]\n",
    "        testSet = self.tdf.iloc[int(self.split*self.tdf.shape[0]):]\n",
    "        trainXW,trainXU,trainY = self.getX1X2Y(trainSet)\n",
    "        testXW,testXU,testY = self.getX1X2Y(testSet)\n",
    "        train_dataset = tf.data.Dataset.from_tensor_slices(\\\n",
    "            (trainXW,trainXU,trainY))\n",
    "        self.train_dataset = train_dataset.shuffle(buffer_size=1024)\\\n",
    "                    .batch(self.batch_size)\n",
    "        val_dataset = tf.data.Dataset.from_tensor_slices(\\\n",
    "            (testXW,testXU,testY))\n",
    "        self.val_dataset = val_dataset.batch(self.batch_size)\n",
    "    def getModel(self):\n",
    "        n1 = tf.keras.Input(shape=(1,))\n",
    "        m1 = tf.keras.Input(shape=(1,))\n",
    "        k1n = tf.keras.layers.Embedding(self.mf.N,self.mf.K)(n1)\n",
    "        k1m = tf.keras.layers.Embedding(self.mf.M,self.mf.K)(m1)\n",
    "        k1n = tf.keras.layers.Flatten()(k1n)\n",
    "        k1m = tf.keras.layers.Flatten()(k1m)\n",
    "        x = tf.keras.layers.Concatenate()([k1n,k1m])\n",
    "        x = tf.keras.layers.Dense(self.hidden)(x)\n",
    "        x = tf.keras.layers.LeakyReLU()(x)\n",
    "        y = tf.keras.layers.Dense(1)(x)\n",
    "        y = tf.reshape(y,[-1])\n",
    "        self.model = tf.keras.Model(inputs=[n1,m1],outputs=[y])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 14903/20000263 [00:01<35:24, 9408.78it/s]\n"
     ]
    }
   ],
   "source": [
    "mf = dnnMF(df,1,batch_size=16,hidden=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0 Train loss: 0.9402\n",
      "Epoch 0 Validation loss: 0.8721\n",
      "Epoch 50 Train loss: 0.0239\n",
      "Epoch 50 Validation loss: 1.0932\n",
      "Epoch 100 Train loss: 0.0124\n",
      "Epoch 100 Validation loss: 1.0414\n",
      "Epoch 150 Train loss: 0.0089\n",
      "Epoch 150 Validation loss: 1.0040\n",
      "Epoch 200 Train loss: 0.0068\n",
      "Epoch 200 Validation loss: 0.9985\n",
      "Epoch 250 Train loss: 0.0061\n",
      "Epoch 250 Validation loss: 0.9990\n",
      "Epoch 300 Train loss: 0.0049\n",
      "Epoch 300 Validation loss: 0.9911\n",
      "Epoch 350 Train loss: 0.0053\n",
      "Epoch 350 Validation loss: 0.9998\n",
      "Epoch 400 Train loss: 0.0047\n",
      "Epoch 400 Validation loss: 0.9915\n",
      "Epoch 450 Train loss: 0.0041\n",
      "Epoch 450 Validation loss: 0.9889\n"
     ]
    }
   ],
   "source": [
    "epochs = 500\n",
    "opt = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
    "printInterval = 50\n",
    "trainLosses = []\n",
    "testLosses = []\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    total_train_loss=0\n",
    "    for i,(x1,x2,y) in enumerate(mf.train_dataset):\n",
    "        with tf.GradientTape() as tape:\n",
    "            pred=mf.model([x1,x2],training=True)\n",
    "            loss = tf.keras.losses.mean_squared_error(y,pred)\n",
    "            total_train_loss+=tf.reduce_sum(loss)\n",
    "        grads = tape.gradient(loss,mf.model.trainable_weights)\n",
    "        opt.apply_gradients(zip(grads,mf.model.trainable_weights))\n",
    "    \n",
    "    if epoch%printInterval==0:\n",
    "        total_val_loss = 0\n",
    "        for j,(x1,x2,y) in enumerate(mf.val_dataset):\n",
    "            pred=mf.model([x1,x2],training=False)\n",
    "            total_val_loss+=tf.reduce_sum(\\\n",
    "                tf.keras.losses.mean_squared_error(y,pred))\n",
    "        print(\"Epoch {} Train loss: {:.4f}\".format(\\\n",
    "                epoch,float(total_train_loss/(i+1))))\n",
    "        print(\"Epoch {} Validation loss: {:.4f}\".format(\\\n",
    "                epoch,float(total_val_loss/(j+1))))\n",
    "        trainLosses.append(total_train_loss/(i+1))\n",
    "        testLosses.append(total_val_loss/(j+1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'loss')"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(range(0,epochs,printInterval),\n",
    "         trainLosses,label='train loss')\n",
    "plt.plot(range(0,epochs,printInterval),\n",
    "         testLosses,label='test loss')\n",
    "plt.legend()\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Movie 2: 3.5 (truth) 3.5541 (pred)\n",
      "Movie 29: 3.5 (truth) 3.6259 (pred)\n",
      "Movie 32: 3.5 (truth) 3.6764 (pred)\n",
      "Movie 47: 3.5 (truth) 3.5630 (pred)\n",
      "Movie 50: 3.5 (truth) 3.5724 (pred)\n",
      "Movie 112: 3.5 (truth) 3.5284 (pred)\n",
      "Movie 151: 4.0 (truth) 4.0249 (pred)\n",
      "Movie 223: 4.0 (truth) 4.0989 (pred)\n",
      "Movie 253: 4.0 (truth) 4.0590 (pred)\n",
      "Movie 260: 4.0 (truth) 4.1127 (pred)\n",
      "Movie 293: 4.0 (truth) 4.0018 (pred)\n",
      "Movie 296: 4.0 (truth) 4.0287 (pred)\n",
      "Movie 318: 4.0 (truth) 4.0681 (pred)\n",
      "Movie 337: 3.5 (truth) 3.5348 (pred)\n",
      "Movie 367: 3.5 (truth) 3.0560 (pred)\n",
      "Movie 541: 4.0 (truth) 4.0268 (pred)\n",
      "Movie 589: 3.5 (truth) 3.5192 (pred)\n",
      "Movie 593: 3.5 (truth) 4.3469 (pred)\n",
      "Movie 653: 3.0 (truth) 3.0294 (pred)\n",
      "Movie 919: 3.5 (truth) 3.5520 (pred)\n",
      "Movie 924: 3.5 (truth) 3.4549 (pred)\n",
      "Movie 1009: 3.5 (truth) 3.5362 (pred)\n",
      "Movie 1036: 4.0 (truth) 4.0122 (pred)\n",
      "Movie 1079: 4.0 (truth) 4.0777 (pred)\n",
      "Movie 1080: 3.5 (truth) 3.9063 (pred)\n",
      "Movie 1089: 3.5 (truth) 3.5341 (pred)\n",
      "Movie 1090: 4.0 (truth) 4.0238 (pred)\n",
      "Movie 1097: 4.0 (truth) 3.9459 (pred)\n",
      "Movie 1136: 3.5 (truth) 3.5055 (pred)\n",
      "Movie 1193: 3.5 (truth) 3.5611 (pred)\n",
      "Movie 1196: 4.5 (truth) 4.5013 (pred)\n",
      "Movie 1198: 4.5 (truth) 4.5501 (pred)\n",
      "Movie 1200: 4.0 (truth) 4.0783 (pred)\n",
      "Movie 1201: 3.0 (truth) 3.0389 (pred)\n",
      "Movie 1208: 3.5 (truth) 3.5192 (pred)\n",
      "Movie 1214: 4.0 (truth) 3.6634 (pred)\n",
      "Movie 1215: 4.0 (truth) 3.1439 (pred)\n",
      "Movie 1217: 3.5 (truth) 3.5231 (pred)\n",
      "Movie 1219: 4.0 (truth) 4.0187 (pred)\n",
      "Movie 1222: 3.5 (truth) 3.5323 (pred)\n",
      "Movie 1240: 4.0 (truth) 4.0397 (pred)\n",
      "Movie 1243: 3.0 (truth) 3.0315 (pred)\n",
      "Movie 1246: 3.5 (truth) 3.6222 (pred)\n",
      "Movie 1249: 4.0 (truth) 4.0156 (pred)\n",
      "Movie 1258: 4.0 (truth) 4.0393 (pred)\n",
      "Movie 1259: 4.0 (truth) 4.4469 (pred)\n",
      "Movie 1261: 3.5 (truth) 4.2906 (pred)\n",
      "Movie 1262: 3.5 (truth) 3.5257 (pred)\n",
      "Movie 1266: 4.0 (truth) 4.0147 (pred)\n",
      "Movie 1278: 4.0 (truth) 3.7711 (pred)\n",
      "Movie 1291: 3.5 (truth) 3.5509 (pred)\n",
      "Movie 1304: 3.0 (truth) 3.0437 (pred)\n",
      "Movie 1321: 4.0 (truth) 4.0178 (pred)\n",
      "Movie 1333: 4.0 (truth) 3.5762 (pred)\n",
      "Movie 1348: 3.5 (truth) 3.5400 (pred)\n",
      "Movie 1350: 3.5 (truth) 3.5201 (pred)\n",
      "Movie 1358: 4.0 (truth) 3.9724 (pred)\n",
      "Movie 1370: 3.0 (truth) 3.8593 (pred)\n",
      "Movie 1374: 4.0 (truth) 4.0498 (pred)\n",
      "Movie 1387: 4.0 (truth) 4.0136 (pred)\n",
      "Movie 1525: 3.0 (truth) 3.5010 (pred)\n",
      "Movie 1584: 3.5 (truth) 3.5195 (pred)\n",
      "Movie 1750: 3.5 (truth) 3.5092 (pred)\n",
      "Movie 1848: 3.5 (truth) 3.5249 (pred)\n",
      "Movie 1920: 3.5 (truth) 3.5166 (pred)\n",
      "Movie 1967: 4.0 (truth) 4.0662 (pred)\n",
      "Movie 1994: 3.5 (truth) 3.5342 (pred)\n",
      "Movie 1997: 3.5 (truth) 3.5687 (pred)\n",
      "Movie 2021: 4.0 (truth) 4.0039 (pred)\n",
      "Movie 2100: 4.0 (truth) 4.0050 (pred)\n",
      "Movie 2118: 4.0 (truth) 4.0038 (pred)\n",
      "Movie 2138: 4.0 (truth) 4.0081 (pred)\n",
      "Movie 2140: 4.0 (truth) 4.0927 (pred)\n",
      "Movie 2143: 4.0 (truth) 3.9884 (pred)\n",
      "Movie 2173: 4.0 (truth) 4.0378 (pred)\n",
      "Movie 2174: 4.0 (truth) 3.5944 (pred)\n",
      "Movie 2193: 4.0 (truth) 4.0398 (pred)\n",
      "Movie 2194: 3.5 (truth) 3.5458 (pred)\n",
      "Movie 2253: 3.5 (truth) 3.5604 (pred)\n",
      "Movie 2288: 4.0 (truth) 4.0859 (pred)\n",
      "Movie 2291: 4.0 (truth) 4.0111 (pred)\n",
      "Movie 2542: 4.0 (truth) 4.0786 (pred)\n",
      "Movie 2628: 4.0 (truth) 3.5112 (pred)\n",
      "Movie 2644: 3.5 (truth) 3.5288 (pred)\n",
      "Movie 2648: 3.5 (truth) 3.5302 (pred)\n",
      "Movie 2664: 3.5 (truth) 3.5226 (pred)\n",
      "Movie 2683: 3.5 (truth) 3.5418 (pred)\n",
      "Movie 2692: 3.5 (truth) 3.9351 (pred)\n",
      "Movie 2716: 3.5 (truth) 3.5615 (pred)\n",
      "Movie 2761: 3.0 (truth) 2.9939 (pred)\n",
      "Movie 2762: 4.0 (truth) 4.2189 (pred)\n",
      "Movie 2804: 3.5 (truth) 3.5511 (pred)\n",
      "Movie 2872: 4.0 (truth) 3.4321 (pred)\n",
      "Movie 2918: 3.5 (truth) 4.1788 (pred)\n",
      "Movie 2944: 4.0 (truth) 4.0011 (pred)\n",
      "Movie 2947: 3.5 (truth) 3.5369 (pred)\n",
      "Movie 2959: 4.0 (truth) 4.0283 (pred)\n",
      "Movie 2968: 4.0 (truth) 4.0380 (pred)\n",
      "Movie 3000: 3.5 (truth) 3.5499 (pred)\n",
      "Movie 3030: 3.0 (truth) 3.7857 (pred)\n"
     ]
    }
   ],
   "source": [
    "targetUser,limit = 1,100\n",
    "x1 = np.array([mf.mf.u2id[targetUser]]).reshape(1,1,1)\n",
    "res = []\n",
    "for i,movie in enumerate(mf.mf.allmovies):\n",
    "    x2 = np.array([mf.mf.m2id[movie]]).reshape(1,1,1)\n",
    "    pred=mf.model([x1,x2],training=False)+mf.mf.mean\n",
    "    pred=pred.numpy()[0]\n",
    "    if movie in mf.u2m[targetUser][0]:\n",
    "        if i<limit:\n",
    "            print(\"Movie {}: {} (truth) {:.4f} (pred)\".format(\\\n",
    "            movie,mf.u2m[targetUser][0][movie],\n",
    "            pred))\n",
    "    else:\n",
    "        store = -pred\n",
    "        heappush(res,(store,movie))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(-5.154361, 110),\n",
       " (-5.0715356, 527),\n",
       " (-4.6174808, 175),\n",
       " (-5.0504875, 778),\n",
       " (-4.987379, 590),\n",
       " (-4.2832704, 1084),\n",
       " (-4.4597116, 8910),\n",
       " (-4.6164913, 2028),\n",
       " (-4.480746, 1704),\n",
       " (-4.649173, 1270)]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res[:10]"
   ]
  },
  {
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}