KNN implementation from scratch.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Movie Recommendation System"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Problem Definition:Creating a user-based movie rating prediction system by using KNN and weighted KNN algoritm."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import pandas as pd \n",
    "os.chdir(r\"E:\\Masaüstü\\bbm409-assignment1\") "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "#importing needed data\n",
    "mv=pd.read_csv(\"movies.csv\").copy()\n",
    "rt=pd.read_csv(\"ratings_train.csv\").copy()\n",
    "myTest=pd.read_csv(\"ratings_test_without_gt_new.csv\")\n",
    "myTest2=myTest[myTest[\"rating\"]!=-1]#feature çıkartmak için convert edilmesi gereken test datası \n",
    "myTest=myTest[myTest[\"rating\"]==-1]#predict edilmesi istenen test değerleri"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>userId</th>\n",
       "      <th>movieId</th>\n",
       "      <th>rating</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>title</th>\n",
       "      <th>genres</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>964982703</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>4.0</td>\n",
       "      <td>847434962</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>2.5</td>\n",
       "      <td>1510577970</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "      <td>4.5</td>\n",
       "      <td>1305696483</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1455209816</td>\n",
       "      <td>Toy Story (1995)</td>\n",
       "      <td>Adventure|Animation|Children|Comedy|Fantasy</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   userId  movieId  rating   timestamp             title  \\\n",
       "0       1        1     4.0   964982703  Toy Story (1995)   \n",
       "1       5        1     4.0   847434962  Toy Story (1995)   \n",
       "2      15        1     2.5  1510577970  Toy Story (1995)   \n",
       "3      17        1     4.5  1305696483  Toy Story (1995)   \n",
       "4      18        1     3.5  1455209816  Toy Story (1995)   \n",
       "\n",
       "                                        genres  \n",
       "0  Adventure|Animation|Children|Comedy|Fantasy  \n",
       "1  Adventure|Animation|Children|Comedy|Fantasy  \n",
       "2  Adventure|Animation|Children|Comedy|Fantasy  \n",
       "3  Adventure|Animation|Children|Comedy|Fantasy  \n",
       "4  Adventure|Animation|Children|Comedy|Fantasy  "
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "###### combining movies.csv and ratings_train.csv with respect to movieıD\n",
    "#combined data\n",
    "new=pd.merge(rt, mv, on='movieId', how='inner')\n",
    "new=pd.merge(rt, mv, on='movieId', how='inner')\n",
    "new.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "#this method creates feature matrix for training data, returns training Data's feature matrix and trainingDataUsers matrix \n",
    "#parameters are combined data and movies.csv. Use only for training data\n",
    "def dataConverter(new,mv):\n",
    "    \n",
    "    new[\"isAd\"]=[1 if \"Adventure\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isAn\"]=[1 if \"Animation\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isCh\"]=[1 if \"Children\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isCo\"]=[1 if \"Comedy\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isFa\"]=[1 if \"Fantasy\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isRo\"]=[1 if \"Romance\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isHo\"]=[1 if \"Horror\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isAc\"]=[1 if \"Action\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isDr\"]=[1 if \"Drama\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isCr\"]=[1 if \"Crime\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isTh\"]=[1 if \"Thriller\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isMy\"]=[1 if \"Mystery\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isSc\"]=[1 if \"Sci-Fi\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isWa\"]=[1 if \"War\" in i else 0 for i in new[\"genres\"]]\n",
    "    new[\"isDo\"]=[1 if \"Documentary\" in i else 0 for i in new[\"genres\"]]\n",
    "    new2=new[new.columns[6:]].copy()\n",
    "    new2[\"userId\"]=new[\"userId\"]\n",
    "    trainingData=new2.groupby('userId').sum()\n",
    "    trainingDataUsers=trainingData.index.to_numpy()\n",
    "    trainingData=trainingData.to_numpy()\n",
    "    return trainingData,trainingDataUsers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "#this method creates feature matrix for test data, returns test Data's feature matrix and test DataUsers matrix \n",
    "#parameters are convert edilmesi gereken test datası(myTest2) and movies.csv. Use only for training data\n",
    "def testDataConverter(new1,mv):\n",
    "    new=new1[new1[\"rating\"]!=-1]\n",
    "    new1=new1[new1[\"rating\"]==-1]\n",
    "    new=pd.merge(new, mv, on='movieId', how='inner')\n",
    "    new[\"isAd\"]=[1 if \"Adventure\" in i else 0 for i in new.genres]\n",
    "    new[\"isAn\"]=[1 if \"Animation\" in i else 0 for i in new.genres]\n",
    "    new[\"isCh\"]=[1 if \"Children\" in i else 0 for i in new.genres]\n",
    "    new[\"isCo\"]=[1 if \"Comedy\" in i else 0 for i in new.genres]\n",
    "    new[\"isFa\"]=[1 if \"Fantasy\" in i else 0 for i in new.genres]\n",
    "    new[\"isRo\"]=[1 if \"Romance\" in i else 0 for i in new.genres]\n",
    "    new[\"isHo\"]=[1 if \"Horror\" in i else 0 for i in new.genres]\n",
    "    new[\"isAc\"]=[1 if \"Action\" in i else 0 for i in new.genres]\n",
    "    new[\"isDr\"]=[1 if \"Drama\" in i else 0 for i in new.genres]\n",
    "    new[\"isCr\"]=[1 if \"Crime\" in i else 0 for i in new.genres]\n",
    "    new[\"isTh\"]=[1 if \"Thriller\" in i else 0 for i in new.genres]\n",
    "    new[\"isMy\"]=[1 if \"Mystery\" in i else 0 for i in new.genres]\n",
    "    new[\"isSc\"]=[1 if \"Sci-Fi\" in i else 0 for i in new.genres]\n",
    "    new[\"isWa\"]=[1 if \"War\" in i else 0 for i in new.genres]\n",
    "    new[\"isDo\"]=[1 if \"Documentary\" in i else 0 for i in new.genres]\n",
    "    new2=new[new.columns[5:]].copy()\n",
    "    new2[\"userId\"]=new[\"userId\"]\n",
    "    testData=new2.groupby('userId').sum()\n",
    "    userIDVector=testData.index.to_numpy()\n",
    "    testData=testData.to_numpy()\n",
    "    return testData,userIDVector\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "#this method create and return a dataframe thats colums are moviesIds and rows are usersIds.Parameter is combined data.\n",
    "#This basicly our known target\n",
    "def trainTargetConverter(new):\n",
    "    dic={}\n",
    "    for i in new.index:\n",
    "        if(new[\"userId\"][i]  not in dic.keys()):\n",
    "            dic[new[\"userId\"][i]]={new[\"movieId\"][i]:new[\"rating\"][i]}\n",
    "        else:\n",
    "              dic[new[\"userId\"][i]].update({new[\"movieId\"][i]:new[\"rating\"][i]})\n",
    "\n",
    "\n",
    "    df = pd.DataFrame(dic) \n",
    "    df=df.T\n",
    "    df=df.apply(lambda row: row.fillna(row.mean()), axis=1)\n",
    "    df=df.sort_index()\n",
    "    return df\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Assigning train trainUser test testUser\n",
    "trainingData,trainingDataUsers=dataConverter(new,mv)\n",
    "testData,testDataUsers=testDataConverter(myTest2,mv)\n",
    "trainTarget=trainTargetConverter(new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Fit and Knn methods\n",
    "\n",
    "from numpy import linalg as LA\n",
    "def fitMethod(train,test,kPara,trainTarget,trainingDataUsers,testDataUsers):\n",
    "    #cosine based similarity used\n",
    "    similarityMatrix=((np.matmul(train, test.T))/(LA.norm(train)*LA.norm(test)))\n",
    "    return KNNMethod(similarityMatrix,trainTarget,kPara,trainingDataUsers,testDataUsers)\n",
    "\n",
    "\n",
    "#This method returns a dataframe that colums are all movies in the movies.csv and rows are testusersId\n",
    "#return predictedValues\n",
    "def KNNMethod(similarityMatrix,trainTarget,kPara,trainingDataUsers,testDataUsers):\n",
    "    sm=similarityMatrix.copy()\n",
    "    sm=pd.DataFrame(sm,index=trainingDataUsers,columns=testDataUsers)\n",
    "    predictValues=[]\n",
    "    dic={}\n",
    "    for i in sm.columns:\n",
    "        col=i\n",
    "        sm=sm.sort_values(by=col, ascending=False)\n",
    "        values=sm.index[0:kPara]\n",
    "        dic[i]={x:(sum(trainTarget.loc[values,x])/len(values)) for x in trainTarget.columns}\n",
    "\n",
    "    predictValues=pd.DataFrame(dic).T\n",
    "   \n",
    "    return predictValues"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "#this method calculates MAE.Parameters ara trainTarget and predictedValues and only for known target\n",
    "def testMethod(trainTarget,predictValues):\n",
    "    sumValue=0\n",
    "    for i in predictValues.index:\n",
    "        for y in predictValues.columns:\n",
    "            sumValue=abs((trainTarget[y][i])-(predictValues[y][i]))+sumValue\n",
    "        \n",
    "    MAE=(1/(len(predictValues)*len(trainTarget.columns)))*sumValue\n",
    "    return MAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "#crossValidation method is only for TESTING, it cant't be used in unknown targets.It returns average of number of error\n",
    "def crossValidation(kFold,trainingData,kPara,trainTarget,trainingDataUsers):#trainingData is data ,k fold\n",
    "    \n",
    "    data=trainingData.copy()\n",
    "    data2=trainingDataUsers.copy()\n",
    "    folds = np.array_split(data, kFold)\n",
    "    foldsUsers=np.array_split(data2,kFold)\n",
    "    listOfAccu=[]\n",
    "    for i in range(kFold):\n",
    "        tempTrain=folds.copy()\n",
    "        tempTrainUser=foldsUsers.copy()\n",
    "        test=folds[i].copy()\n",
    "        testDataUsers=foldsUsers[i].copy()\n",
    "        del tempTrain[i]\n",
    "        del tempTrainUser[i]\n",
    "        train=np.concatenate(tempTrain, axis=0).copy()\n",
    "        trainingDataUsers=np.concatenate(tempTrainUser, axis=0).copy()\n",
    "        predictValues=fitMethod(train,test,kPara,trainTarget,trainingDataUsers,testDataUsers)\n",
    "        listOfAccu.append(testMethod(trainTarget,predictValues))\n",
    "    print(\"MAE for k-fold \",kFold,\" \",sum(listOfAccu)/kFold)  \n",
    "    return sum(listOfAccu)/kFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "#predict method takes userId and Movie tuple or list and predictedValues as parameters return and print userId and rating\n",
    "def predictMethod(userAndMovie,predictedValues):\n",
    "    listOfPrediction=[]\n",
    "    for i in userAndMovie:\n",
    "        print(i[0],\",\",predictedValues[i[1]][i[0]])\n",
    "        listOfPrediction.append([i[0],predictedValues[i[1]][i[0]]])\n",
    "    return listOfprediction "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "#less test data is used in this cell due to the time issues. k parameter is 3\n",
    "predicted=fitMethod(trainingData[20:],trainingData[0:20],3,trainTarget,trainingDataUsers[20:],trainingDataUsers[0:20])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "      <th>...</th>\n",
       "      <th>193565</th>\n",
       "      <th>193567</th>\n",
       "      <th>193571</th>\n",
       "      <th>193573</th>\n",
       "      <th>193579</th>\n",
       "      <th>193581</th>\n",
       "      <th>193583</th>\n",
       "      <th>193585</th>\n",
       "      <th>193587</th>\n",
       "      <th>193609</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.666667</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.966319</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>2.04735</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>3.013669</td>\n",
       "      <td>2.763638</td>\n",
       "      <td>3.299652</td>\n",
       "      <td>...</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>4.000000</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.54735</td>\n",
       "      <td>3.449124</td>\n",
       "      <td>2.782457</td>\n",
       "      <td>2.829807</td>\n",
       "      <td>2.579776</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>...</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>3.666667</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.966319</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>2.04735</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>3.013669</td>\n",
       "      <td>2.763638</td>\n",
       "      <td>3.299652</td>\n",
       "      <td>...</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>3.666667</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.966319</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>2.04735</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>3.013669</td>\n",
       "      <td>2.763638</td>\n",
       "      <td>3.299652</td>\n",
       "      <td>...</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "      <td>3.144321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>4.000000</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.833333</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.54735</td>\n",
       "      <td>3.449124</td>\n",
       "      <td>2.782457</td>\n",
       "      <td>2.829807</td>\n",
       "      <td>2.579776</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>...</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "      <td>2.960459</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 8904 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      1         2         3         4        5         6         7       \\\n",
       "4   3.666667  2.833333  2.966319  3.144321  2.04735  3.500000  2.833333   \n",
       "7   4.000000  2.833333  2.833333  2.960459  2.54735  3.449124  2.782457   \n",
       "10  3.666667  2.833333  2.966319  3.144321  2.04735  3.500000  2.833333   \n",
       "16  3.666667  2.833333  2.966319  3.144321  2.04735  3.500000  2.833333   \n",
       "19  4.000000  2.833333  2.833333  2.960459  2.54735  3.449124  2.782457   \n",
       "\n",
       "      8         9         10      ...    193565    193567    193571    193573  \\\n",
       "4   3.013669  2.763638  3.299652  ...  3.144321  3.144321  3.144321  3.144321   \n",
       "7   2.829807  2.579776  3.500000  ...  2.960459  2.960459  2.960459  2.960459   \n",
       "10  3.013669  2.763638  3.299652  ...  3.144321  3.144321  3.144321  3.144321   \n",
       "16  3.013669  2.763638  3.299652  ...  3.144321  3.144321  3.144321  3.144321   \n",
       "19  2.829807  2.579776  3.500000  ...  2.960459  2.960459  2.960459  2.960459   \n",
       "\n",
       "      193579    193581    193583    193585    193587    193609  \n",
       "4   3.144321  3.144321  3.144321  3.144321  3.144321  3.144321  \n",
       "7   2.960459  2.960459  2.960459  2.960459  2.960459  2.960459  \n",
       "10  3.144321  3.144321  3.144321  3.144321  3.144321  3.144321  \n",
       "16  3.144321  3.144321  3.144321  3.144321  3.144321  3.144321  \n",
       "19  2.960459  2.960459  2.960459  2.960459  2.960459  2.960459  \n",
       "\n",
       "[5 rows x 8904 columns]"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predicted.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.718745701427801\n"
     ]
    }
   ],
   "source": [
    "#MAE for k hyperparameter =3\n",
    "MAE=testMethod(trainTarget,predicted)\n",
    "print(MAE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE for  3  hyperparameter 0.718745701427801\n",
      "MAE for  4  hyperparameter 0.7062708369633199\n",
      "MAE for  5  hyperparameter 0.6368167278080952\n",
      "MAE for  6  hyperparameter 0.6030906380597183\n",
      "MAE for  7  hyperparameter 0.557272431223616\n",
      "MAE for  8  hyperparameter 0.5363309773322302\n",
      "MAE for  9  hyperparameter 0.5261050499817567\n",
      "MAE for  10  hyperparameter 0.5329945908934768\n"
     ]
    }
   ],
   "source": [
    "#trying different k parameter\n",
    "for i in range(3,11):\n",
    "    predicted=fitMethod(trainingData[20:],trainingData[0:20],i,trainTarget,trainingDataUsers[20:],trainingDataUsers[0:20])\n",
    "    MAE=testMethod(trainTarget,predicted)\n",
    "    print(\"MAE for \",i,\" hyperparameter\",MAE)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE for k-fold  5   0.4845092106479411\n"
     ]
    }
   ],
   "source": [
    "#testing algoritm accuracy with cross validation kFold is 5 k parameter is 9.\n",
    "#less training data is used in this test due to the time issue\n",
    "predictedCv=crossValidation(5,trainingData[0:50],9,trainTarget,trainingDataUsers[0:50])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE for k-fold  5   0.48406924613027585\n"
     ]
    }
   ],
   "source": [
    "#for k=10 hyperparameter\n",
    "predictedCv=crossValidation(5,trainingData[0:50],10,trainTarget,trainingDataUsers[0:50])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "#weightedFitMethod and WeightedKnn method takes same parameters and return same values like normal fit and knn method.\n",
    "#but in this knnMethod neighbors weight is like below\n",
    "# (k is hyperparameter k) 1.neighbor*k+2.neighbor*k-1+....+k.neighbor*1\n",
    "def weightedFitMethod(train,test,kPara,trainTarget,trainingDataUsers,testDataUsers):\n",
    "    similarityMatrix=((np.matmul(train, test.T))/(LA.norm(train)*LA.norm(test)))\n",
    "    return weightedKNNMethod(similarityMatrix,trainTarget,kPara,trainingDataUsers,testDataUsers)\n",
    "    \n",
    "def weightedKNNMethod(similarityMatrix,trainTarget,kPara,trainingDataUsers,testDataUsers):\n",
    "    sm=similarityMatrix.copy()\n",
    "    sm=pd.DataFrame(sm,index=trainingDataUsers,columns=testDataUsers)\n",
    "    predictValues=[]\n",
    "    dic={}\n",
    "    kParaList=[k for k in range(kPara,0,-1)]\n",
    "    sumKPara=sum(kParaList)\n",
    "    for i in sm.columns:\n",
    "        col=i\n",
    "        sm=sm.sort_values(by=col, ascending=False)\n",
    "        values=sm.index[0:kPara]\n",
    "        mapped=zip(values,kParaList)\n",
    "        mapped=[list(i) for i in mapped]\n",
    "        dictionary={x[0]:x[1] for x in mapped}\n",
    "        \n",
    "        \n",
    "        copyTrainTarget=trainTarget.loc[values,:].copy()\n",
    "        \n",
    "        copyTrainTarget=copyTrainTarget.mul(kParaList, axis=0)\n",
    "            \n",
    "        dic[i]={x:(sum(copyTrainTarget.loc[values,x])/sumKPara) for x in trainTarget.columns}\n",
    "        \n",
    "    \n",
    "    predictValues=pd.DataFrame(dic).T\n",
    "   \n",
    "    return predictValues\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE for  3  hyperparameter 0.6990316897214891  in weighted KNN\n",
      "MAE for  4  hyperparameter 0.6996309669354022  in weighted KNN\n",
      "MAE for  5  hyperparameter 0.6771204180778738  in weighted KNN\n",
      "MAE for  6  hyperparameter 0.6549371969429224  in weighted KNN\n",
      "MAE for  7  hyperparameter 0.6292831653646285  in weighted KNN\n",
      "MAE for  8  hyperparameter 0.6068984779981725  in weighted KNN\n",
      "MAE for  9  hyperparameter 0.5895159628926648  in weighted KNN\n",
      "MAE for  10  hyperparameter 0.5780313240983389  in weighted KNN\n"
     ]
    }
   ],
   "source": [
    "for i in range(3,11):\n",
    "    predictedWe=weightedFitMethod(trainingData[20:],trainingData[0:20],i,trainTarget,trainingDataUsers[20:],trainingDataUsers[0:20])\n",
    "    MAE=testMethod(trainTarget,predictedWe)\n",
    "    print(\"MAE for \",i,\" hyperparameter\",MAE,\" in weighted KNN\")"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}