Recommendation Systems and Truncated SVD SGD algorithm to predict ratings.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "YeM0ZBWUVtXR"
   },
   "source": [
    "# <font color='red'>SGD Algorithm to predict movie ratings</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "AL6njTf8WBO0"
   },
   "source": [
    "<pre>\n",
    "1. Download the data from <a href='https://drive.google.com/open?id=1-1z7iDB52cB6_JpO7Dqa-eOYSs-mivpq'> here </a>\n",
    "2. The data will be of this format, each data point is represented as a triplet of user_id, movie_id and rating \n",
    "<table>\n",
    "<tr><th>user_id</th><th>movie_id</th><th>rating</th></tr>\n",
    "<tr><td>77</td><td>236</td><td>3</td></tr>\n",
    "<tr><td>471</td><td>208</td><td>5</td></tr>\n",
    "<tr><td>641</td><td>401</td><td>4</td></tr>\n",
    "<tr><td>31</td><td>298</td><td>4</td></tr>\n",
    "<tr><td>58</td><td>504</td><td>5</td></tr>\n",
    "<tr><td>235</td><td>727</td><td>5</td></tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "73dhFsT0WSSB"
   },
   "source": [
    "## <font color='red'>Task 1</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HY0frS6EWaEV"
   },
   "source": [
    "<font color='red'><b>Predict the rating for a given (user_id, movie_id) pair </b> </font>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "R-ISYxaVbT8L"
   },
   "source": [
    "Predicted rating $\\hat{y}_{ij}$ for user i, movied j pair is calcuated as $\\hat{y}_{ij} = \\mu + b_i + c_j + u_i^T v_j$ , here we will be finding the best values of $b_{i}$ and $c_{j}$ using SGD algorithm with the optimization problem for N users and M movies is defined as"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8Aj8SXeQWlZd"
   },
   "source": [
    "$$\n",
    "L = \\min_{ b, c, \\{ u_i \\}_{i=1}^N, \\{ v_j \\}_{j=1}^M}\n",
    "\\quad\n",
    "\\alpha \\Big(\n",
    "    \\sum_{j} \\sum_{k} v_{jk}^2 \n",
    "    + \\sum_{i} \\sum_{k} u_{ik}^2 \n",
    "    + \\sum_{i} b_i^2\n",
    "    + \\sum_{j} c_i^2\n",
    "    \\Big)\n",
    "+ \\sum_{i,j \\in \\mathcal{I}^{\\text{train}}}\n",
    "    (y_{ij} - \\mu - b_i - c_j - u_i^T v_j)^2\n",
    "$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2Q5bnWyZXrM7"
   },
   "source": [
    "<ul>\n",
    "<li><span class=\"math\">\\(\\mu\\)</span> : scalar mean rating</li>\n",
    "<li><span class=\"math\">\\(b_i\\)</span> : scalar bias term for user <span class=\"math\">\\(i\\)</span></li>\n",
    "<li><span class=\"math\">\\(c_j\\)</span> : scalar bias term for movie <span class=\"math\">\\(j\\)</span></li>\n",
    "<li><span class=\"math\">\\(u_i\\)</span> : K-dimensional vector for user <span class=\"math\">\\(i\\)</span></li>\n",
    "<li><span class=\"math\">\\(v_j\\)</span> : K-dimensional vector for movie <span class=\"math\">\\(j\\)</span></li>\n",
    "</ul>\n",
    "\n",
    " $ \\ $\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ZWQyB5hfy3u7"
   },
   "source": [
    "1. Construct adjacency matrix with the given data, assuming its graph and the weight of each edge is the rating given by user to the movie\n",
    "\n",
    "<img src='https://i.imgur.com/rmUCGMb.jpg' width=200>\n",
    "\n",
    "   you can construct this matrix like $A[i][j]=r_{ij}$ here $i$ is user_id, $j$ is movie_id and $r_{ij}$ is rating given by user $i$ to the movie $j$\n",
    "\n",
    "2. We will Apply SVD decomposition on the Adjaceny matrix <a href='https://stackoverflow.com/a/31528944/4084039'>link1</a>, <a href='https://machinelearningmastery.com/singular-value-decomposition-for-machine-learning/'> link2</a> and get three matrices $U, \\sum, V$ such that $U \\times \\sum \\times V^T = A$, <br> \n",
    "if $A$ is of dimensions $N \\times M$ then <br>\n",
    "U is of $N \\times k$, <br>\n",
    "$\\sum$ is of $k \\times k$ and <br>\n",
    "$V$ is $M \\times k$ dimensions. <br>\n",
    "\n",
    "   *.  So the matrix $U$ can be represented as matrix representation of users, where each row $u_{i}$ represents a k-dimensional vector for a user\n",
    "\n",
    "   *. So the matrix $V$ can be represented as matrix representation of movies, where each row $v_{j}$ represents a k-dimensional vector for a movie.\n",
    "3. Compute $\\mu$ , $\\mu$  represents the mean of all the rating given in the dataset.(write your code in <font color='blue'>def m_u()</font>)\n",
    "4. For each unique user initilize a bias value $B_{i}$ to zero, so if we have $N$ users $B$ will be a $N$ dimensional vector, the $i^{th}$ value of the $B$ will corresponds to the bias term for $i^{th}$ user (write your code in <font color='blue'>def initialize()</font>)\n",
    "\n",
    "5. For each unique movie initilize a bias value $C_{j}$ zero, so if we have $M$ movies $C$ will be a $M$ dimensional vector, the $j^{th}$ value of the $C$ will corresponds to the bias term for $j^{th}$ movie (write your code in <font color='blue'>def initialize()</font>)\n",
    "\n",
    "6. Compute dL/db_i (Write you code in <font color='blue'> def derivative_db()</font>)\n",
    "7. Compute dL/dc_j(write your code in <font color='blue'> def derivative_dc()</font>\n",
    "\n",
    "8. Print the mean squared error with predicted ratings.\n",
    "\n",
    "<pre>\n",
    "for each epoch:\n",
    "    for each pair of (user, movie):\n",
    "        b_i =  b_i - learning_rate * dL/db_i\n",
    "        c_j =  c_j - learning_rate * dL/dc_j\n",
    "predict the ratings with formula\n",
    "</pre>\n",
    "$\\hat{y}_{ij} = \\mu + b_i + c_j + \\text{dot_product}(u_i , v_j) $\n",
    "\n",
    "9. you can choose any learning rate and regularization term in the range $10^{-3}  \\text{ to } 10^2$  <br>\n",
    "  \n",
    "10. __bonus__: instead of using SVD decomposition you can learn the vectors $u_i$, $v_j$ with the help of SGD algo similar to $b_i$ and $c_j$ "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "As1XN6K680-h",
    "outputId": "3031e8ce-e134-4c43-98e7-2be88933ae78"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "unbiIm1g_ZLK"
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "YABscNLd9Hk8"
   },
   "outputs": [],
   "source": [
    "import os \n",
    "os.chdir('/content/drive/MyDrive/aaic (1)/RecommendationSystem_TruncatedSVD')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "jMEaJU929OGJ"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3-aBnRepA6gy"
   },
   "source": [
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IP_6xMAZA4mE"
   },
   "source": [
    " # <font color='red'>Task 2 </font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-9HCN_3WA2au"
   },
   "source": [
    "As we know U is the learned matrix of user vectors, with its i-th row as the vector ui for user i. Each row of U can be seen as a \"feature vector\" for a particular user.\n",
    "\n",
    "The question we'd like to investigate is this: do our computed per-user features that are optimized for predicting movie ratings contain anything to do with gender?\n",
    "\n",
    "The provided data file <a href='https://drive.google.com/open?id=1PHFdJh_4gIPiLH5Q4UErH8GK71hTrzlY'>user_info.csv</a> contains an is_male column indicating which users in the dataset are male. Can you predict this signal given the features U?\n",
    "\n",
    "\n",
    "> __Note 1__ : there is no train test split in the data, the goal of this assignment is to give an intution about how to do matrix factorization with the help of SGD and application of truncated SVD. for better understanding of the collabarative fillerting please check netflix case study. <br><br>\n",
    "> __Note 2__ : Check if scaling of $U$, $V$ matrices improve the metric "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ovFCo1JCBIXM"
   },
   "source": [
    "<br>\n",
    "\n",
    "\n",
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "iVJo-3njBQLf"
   },
   "source": [
    "<font color='red'> Reading the csv file </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 206
    },
    "id": "hEhmfRD637EW",
    "outputId": "2dbe3855-17a5-4080-d320-65bd9f7aa95c"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-74860854-3e4c-462d-8135-df6f017ab0a3\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>user_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>772</td>\n",
       "      <td>36</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>471</td>\n",
       "      <td>228</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>641</td>\n",
       "      <td>401</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>312</td>\n",
       "      <td>98</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>58</td>\n",
       "      <td>504</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-74860854-3e4c-462d-8135-df6f017ab0a3')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-74860854-3e4c-462d-8135-df6f017ab0a3 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-74860854-3e4c-462d-8135-df6f017ab0a3');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "   user_id  item_id  rating\n",
       "0      772       36       3\n",
       "1      471      228       5\n",
       "2      641      401       4\n",
       "3      312       98       4\n",
       "4       58      504       5"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "data=pd.read_csv('ratings_train.csv')\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "66ibGJ74hCde",
    "outputId": "f2184e83-4479-4fef-9cf2-795746624b08"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(89992, 3)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mvB8SDS_hW03"
   },
   "source": [
    "<font color='red'>Create your adjacency matrix </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "t44MNT40hZQW"
   },
   "outputs": [],
   "source": [
    "from scipy.sparse import csr_matrix\n",
    "adjacency_matrix =csr_matrix((data.rating.values, (data.user_id.values,\n",
    "                                              data.item_id.values)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "-mCgC0WbhZTO",
    "outputId": "dc6468f0-955d-4ebf-8bd1-d9e99595da80"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(943, 1681)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adjacency_matrix.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xWxWamQUCO-6",
    "outputId": "322cfef2-9e00-44a5-8574-8c937891dfc8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[5, 0, 4, ..., 0, 0, 0],\n",
       "       [4, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       ...,\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 0, 0, ..., 0, 0, 0],\n",
       "       [0, 5, 0, ..., 0, 0, 0]], dtype=int64)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adjacency_matrix.toarray()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "U7VwkRNeHpWE"
   },
   "source": [
    "**The unique items in the given csv file are 1662 only . But the id's vary from 0-1681 but they are not continuous and hence \n",
    "you'll get matrix of size 943x1681.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gXDf1RCUBsYN"
   },
   "source": [
    "<font color='red'> SVD decompostion</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OJPWI9VwD_ih"
   },
   "source": [
    "Sample code for SVD decompostion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "GATD35bmBszc",
    "outputId": "7506169f-4262-4a55-8972-26d8f4fa69e2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(20, 5)\n",
      "(5,)\n",
      "(10, 5)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.utils.extmath import randomized_svd\n",
    "import numpy as np \n",
    "matrix = np.random.random((20, 10))\n",
    "U, Sigma, VT = randomized_svd(matrix, n_components=5,n_iter=5, random_state=None)\n",
    "print(U.shape)\n",
    "print(Sigma.shape)\n",
    "print(VT.T.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ePDgwALQEJoB"
   },
   "source": [
    "<font color='red'>code for SVD decompostion</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ZYnsKBmFEIg3",
    "outputId": "ae6ab664-753f-419c-cf47-79c4d043ea2a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(943, 50)\n",
      "(50,)\n",
      "(1681, 50)\n"
     ]
    }
   ],
   "source": [
    "\n",
    "from sklearn.utils.extmath import randomized_svd\n",
    "import numpy as np \n",
    "U, Sigma, VT = randomized_svd(adjacency_matrix, n_components=50,n_iter=5, random_state=None)\n",
    "print(U.shape)\n",
    "print(Sigma.shape)\n",
    "print(VT.T.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "83Vh4NoO_JyU"
   },
   "source": [
    "<font color='red'>Compute mean of ratings</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "cBHuCn2QSEnl"
   },
   "outputs": [],
   "source": [
    "def m_u(ratings):\n",
    "    '''In this function, we will compute mean for all the ratings'''\n",
    "    return ratings.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "iu1nn-1x3ebp",
    "outputId": "c6a0c42a-d18a-4e9e-bb89-0b1e416f91dc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.529480398257623\n"
     ]
    }
   ],
   "source": [
    "mu=m_u(data['rating'])\n",
    "print(mu)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qSvAW1X94g3G"
   },
   "source": [
    "<font color='red'>Initialize $B_{i}$ and $C_{j}$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "AyEJqPka4lBW"
   },
   "outputs": [],
   "source": [
    "def initialize(dim):\n",
    "    '''In this function, we will initialize bias value 'B' and 'C'.'''\n",
    "    return np.zeros(dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "Nlae9QAQ43Xz"
   },
   "outputs": [],
   "source": [
    "dim=adjacency_matrix.shape[0] # give the number of dimensions for b_i (Here b_i corresponds to users)\n",
    "b_i=initialize(dim)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "rwuopn4HoEbP"
   },
   "outputs": [],
   "source": [
    "dim=adjacency_matrix.shape[1] # give the number of dimensions for c_j (Here c_j corresponds to movies)\n",
    "c_j=initialize(dim)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DTDK4ZR18MrZ"
   },
   "source": [
    "<font color='red'>Compute dL/db_i</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BCGzxaR1RcNw"
   },
   "source": [
    "$$\n",
    "L = \\min_{ b, c, \\{ u_i \\}_{i=1}^N, \\{ v_j \\}_{j=1}^M}\n",
    "\\quad\n",
    "\\alpha \\Big(\n",
    "    \\sum_{j} \\sum_{k} v_{jk}^2 \n",
    "    + \\sum_{i} \\sum_{k} u_{ik}^2 \n",
    "    + \\sum_{i} b_i^2\n",
    "    + \\sum_{j} c_i^2\n",
    "    \\Big)\n",
    "+ \\sum_{i,j \\in \\mathcal{I}^{\\text{train}}}\n",
    "    (y_{ij} - \\mu - b_i - c_j - u_i^T v_j)^2\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "3NFzVC1N8S4L"
   },
   "outputs": [],
   "source": [
    "def derivative_db(user_id,item_id,rating,U,V,mu,alpha,b_i):\n",
    "    '''In this function, we will compute dL/db_i'''\n",
    "    regularization=2*alpha*b_i[user_id]\n",
    "    loss=-2*(rating-mu-b_i[user_id]-c_j[item_id]-np.dot(U[user_id],V.T[item_id]))\n",
    "    total_loss=loss+regularization\n",
    "    return total_loss"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1Kp0hC_b9v60"
   },
   "source": [
    "<font color='red'>Compute dL/dc_j</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "FAtSYMrc9UqJ"
   },
   "outputs": [],
   "source": [
    "def derivative_dc(user_id,item_id,rating,U,V,mu, alpha,c_j):\n",
    "    '''In this function, we will compute dL/dc_j'''\n",
    "\n",
    "    return alpha*2*c_j-2*(rating-mu-c_j-np.dot(U[user_id],V.T[item_id]))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "lg5XNbDWCIKI"
   },
   "source": [
    "<font color='red'>Compute MSE (mean squared error) for predicted ratings</font>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7WUjNy0TDQX6"
   },
   "source": [
    "for each epoch, print the MSE value"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "D2pCy1AKCafw"
   },
   "source": [
    "<pre>\n",
    "for each epoch:\n",
    "\n",
    "    for each pair of (user, movie):\n",
    "\n",
    "        b_i =  b_i - learning_rate * dL/db_i\n",
    "\n",
    "        c_j =  c_j - learning_rate * dL/dc_j\n",
    "\n",
    "predict the ratings with formula\n",
    "</pre>\n",
    "\n",
    "$\\hat{y}_{ij} = \\mu + b_i + c_j + \\text{dot_product}(u_i , v_j) $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 424
    },
    "id": "XfcxWnm6u44V",
    "outputId": "8ba196a7-f5c7-45a8-e2ff-c26c2395eba4"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-7daa75e7-c1f2-42a5-8cc4-1025452c1f0f\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>user_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>rating</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>772</td>\n",
       "      <td>36</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>471</td>\n",
       "      <td>228</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>641</td>\n",
       "      <td>401</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>312</td>\n",
       "      <td>98</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>58</td>\n",
       "      <td>504</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89987</th>\n",
       "      <td>415</td>\n",
       "      <td>813</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89988</th>\n",
       "      <td>842</td>\n",
       "      <td>120</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89989</th>\n",
       "      <td>574</td>\n",
       "      <td>505</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89990</th>\n",
       "      <td>757</td>\n",
       "      <td>472</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89991</th>\n",
       "      <td>503</td>\n",
       "      <td>204</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>89992 rows × 3 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-7daa75e7-c1f2-42a5-8cc4-1025452c1f0f')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-7daa75e7-c1f2-42a5-8cc4-1025452c1f0f button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-7daa75e7-c1f2-42a5-8cc4-1025452c1f0f');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "       user_id  item_id  rating\n",
       "0          772       36       3\n",
       "1          471      228       5\n",
       "2          641      401       4\n",
       "3          312       98       4\n",
       "4           58      504       5\n",
       "...        ...      ...     ...\n",
       "89987      415      813       4\n",
       "89988      842      120       3\n",
       "89989      574      505       2\n",
       "89990      757      472       5\n",
       "89991      503      204       3\n",
       "\n",
       "[89992 rows x 3 columns]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "kiKj-M6d2a-Y",
    "outputId": "d86e876f-8271-4069-9134-c3d2e1ddfec8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 1: loss = 0.883469065322984\n",
      "epoch 2: loss = 0.858799422302942\n",
      "epoch 3: loss = 0.8498069956234285\n",
      "epoch 4: loss = 0.8454761037335727\n",
      "epoch 5: loss = 0.843063003201758\n",
      "epoch 6: loss = 0.8415841589046563\n",
      "epoch 7: loss = 0.8406139323077432\n",
      "epoch 8: loss = 0.8399441488693551\n",
      "epoch 9: loss = 0.8394635226400652\n",
      "epoch 10: loss = 0.8391082517223324\n",
      "epoch 11: loss = 0.8388395875383562\n",
      "epoch 12: loss = 0.8386328274992275\n",
      "epoch 13: loss = 0.8384715627377638\n",
      "epoch 14: loss = 0.8383445070607498\n",
      "epoch 15: loss = 0.8382436651922268\n",
      "epoch 16: loss = 0.838163229840099\n",
      "epoch 17: loss = 0.8380988925672792\n",
      "epoch 18: loss = 0.8380473987628964\n",
      "epoch 19: loss = 0.8380062516745437\n",
      "epoch 20: loss = 0.8379735103625262\n",
      "epoch 21: loss = 0.83794764853522\n",
      "epoch 22: loss = 0.8379274538720978\n",
      "epoch 23: loss = 0.8379119549019296\n",
      "epoch 24: loss = 0.8379003670296093\n",
      "epoch 25: loss = 0.8378920521221501\n",
      "epoch 26: loss = 0.837886487859948\n",
      "epoch 27: loss = 0.8378832442291251\n",
      "epoch 28: loss = 0.8378819653083074\n",
      "epoch 29: loss = 0.8378823550297175\n",
      "epoch 30: loss = 0.8378841659571852\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import mean_squared_error\n",
    "learning_rate=0.01\n",
    "alpha=0.01\n",
    "epochs=30\n",
    "prediction=[]\n",
    "mse=[]\n",
    "y_actual=data['rating']\n",
    "for epoch in range(1,epochs+1):\n",
    "  prediction=[]\n",
    "  for user_id,item_id,rating in zip(data['user_id'],data['item_id'],data['rating']):\n",
    "    d_b=derivative_db(user_id,item_id,rating,U,VT,mu,alpha,b_i)\n",
    "    b_i[user_id]=b_i[user_id]-learning_rate*d_b\n",
    "    d_c=derivative_dc(user_id, item_id, rating, U, VT, mu, alpha, c_j[item_id])\n",
    "    c_j[item_id]=c_j[item_id]-learning_rate*d_c\n",
    "  for user_id,item_id,rating in zip(data['user_id'],data['item_id'],data['rating']):\n",
    "    pred=mu+b_i[user_id]+c_j[item_id]+np.dot(U[user_id],VT.T[item_id])\n",
    "    prediction.append(pred)\n",
    "  mse.append(mean_squared_error(y_actual,prediction))\n",
    "  print(f\"epoch {epoch}: loss = {mse[-1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "1qzxZKAf3Cku",
    "outputId": "d080fe41-fb2d-4188-e1af-d73d1e29b685"
   },
   "outputs": [
    {
     "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": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(mse,label='mean squared error (loss)')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('MSE')\n",
    "plt.title('epoch vs MSE')\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "NeWAGkT6C9kq"
   },
   "source": [
    "<br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "lkIQOOo1C9o7"
   },
   "source": [
    "# <font color='red'> Task 2</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9kl4Ryi_7E_T"
   },
   "source": [
    "- For this task we have to consider the user_matrix U and the user_info.csv file.\n",
    "- consider is_male columns as  output features and rest as input features. fit a model by posing this problem as binary classification task.\n",
    "- apply any model like Logistic regression or Decision tree and check the performance of the model. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "id": "syygjRvwTJn-"
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import SGDClassifier\n",
    "from sklearn.model_selection import GridSearchCV,train_test_split\n",
    "from sklearn.metrics import confusion_matrix,roc_auc_score,roc_curve,auc\n",
    "from seaborn import heatmap"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "id": "c1FTc39gDdti"
   },
   "outputs": [],
   "source": [
    "user_data=pd.read_csv('user_info.csv.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 424
    },
    "id": "s7e_3BBsHpWO",
    "outputId": "7e525eba-a478-48b8-edd6-8d1d7227ac87"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-92163778-223f-4f18-a6ea-9d4340cdc74a\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <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>user_id</th>\n",
       "      <th>age</th>\n",
       "      <th>is_male</th>\n",
       "      <th>orig_user_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>24</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>53</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>24</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>33</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>938</th>\n",
       "      <td>938</td>\n",
       "      <td>26</td>\n",
       "      <td>0</td>\n",
       "      <td>939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>939</th>\n",
       "      <td>939</td>\n",
       "      <td>32</td>\n",
       "      <td>1</td>\n",
       "      <td>940</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>940</th>\n",
       "      <td>940</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>941</th>\n",
       "      <td>941</td>\n",
       "      <td>48</td>\n",
       "      <td>0</td>\n",
       "      <td>942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>942</th>\n",
       "      <td>942</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "      <td>943</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>943 rows × 4 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-92163778-223f-4f18-a6ea-9d4340cdc74a')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-92163778-223f-4f18-a6ea-9d4340cdc74a button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-92163778-223f-4f18-a6ea-9d4340cdc74a');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "     user_id  age  is_male  orig_user_id\n",
       "0          0   24        1             1\n",
       "1          1   53        0             2\n",
       "2          2   23        1             3\n",
       "3          3   24        1             4\n",
       "4          4   33        0             5\n",
       "..       ...  ...      ...           ...\n",
       "938      938   26        0           939\n",
       "939      939   32        1           940\n",
       "940      940   20        1           941\n",
       "941      941   48        0           942\n",
       "942      942   22        1           943\n",
       "\n",
       "[943 rows x 4 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "id": "M3K5ZHmSHpWO"
   },
   "outputs": [],
   "source": [
    "X=U\n",
    "y=user_data['is_male'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "id": "V5wg8BHWSbwF"
   },
   "outputs": [],
   "source": [
    "X_train,X_test,y_train,y_test=train_test_split(X,y,stratify=y,random_state=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "id": "JVFEDUGbSc7V"
   },
   "outputs": [],
   "source": [
    "model=SGDClassifier(loss='log',random_state=23)\n",
    "paras={'alpha':[0.00001,0.001,0.01,0.1,1,10,100]}\n",
    "clf=GridSearchCV(model,param_grid=paras,scoring='roc_auc',return_train_score=True,cv=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "T7Le5VAISeKN",
    "outputId": "d88cc21d-07d5-4a8e-898a-07b334927024"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=10, estimator=SGDClassifier(loss='log', random_state=23),\n",
       "             param_grid={'alpha': [1e-05, 0.001, 0.01, 0.1, 1, 10, 100]},\n",
       "             return_train_score=True, scoring='roc_auc')"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 295
    },
    "id": "Sqhf9rGUSfNt",
    "outputId": "9360ef6f-871e-4d32-bd66-05e8664135b5"
   },
   "outputs": [
    {
     "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(paras['alpha'],clf.cv_results_['mean_test_score'],label='CV_score')\n",
    "plt.plot(paras['alpha'],clf.cv_results_['mean_train_score'],label='train_score')\n",
    "plt.legend()\n",
    "plt.title('alpha vs AUC')\n",
    "plt.xlabel('aplha')\n",
    "plt.ylabel('AUC_score')\n",
    "plt.xlim(0,100)\n",
    "plt.ylim(0.5,1)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "nsMNQQ0ESgSt",
    "outputId": "357703f5-ff35-4f01-ef0b-976ac56f304b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(alpha=1e-05, loss='log', random_state=23)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_model=clf.best_estimator_\n",
    "best_model.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5QqX7qNxShou",
    "outputId": "57e4bb67-d5bc-4afe-bf88-3720a3f1b1fd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on train_data 0.7468175388967468\n",
      "Accuracy on test_data 0.788135593220339\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy on train_data {}\".format(best_model.score(X_train,y_train))) #imbalanced Dataset\n",
    "print(\"Accuracy on test_data {}\".format(best_model.score(X_test,y_test)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 281
    },
    "id": "cqMEs11DSj0-",
    "outputId": "6d516d92-dc36-43a3-b7ec-3a7f551fc28b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred=best_model.predict(X_train)\n",
    "temp=pd.DataFrame(confusion_matrix(y_train,y_pred).T,index=['Predicted 0','Predicted 1'],columns=['actual 0','actual 1'])\n",
    "heatmap(temp,annot=True,fmt='g')\n",
    "plt.title('Confusion Matrix for X_train (threshold=0.5)')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 281
    },
    "id": "na5uWhOFSk7l",
    "outputId": "9192e826-1d8a-49df-8249-33399ffa2395"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred=best_model.predict(X_test)\n",
    "temp=pd.DataFrame(confusion_matrix(y_test,y_pred).T,index=['Predicted 0','Predicted 1'],columns=['actual 0','actual 1'])\n",
    "heatmap(temp,annot=True,fmt='g')\n",
    "plt.title('Confusion Matrix for X_test (threshold=0.5)')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2t1X8TxDSl_O",
    "outputId": "fe7a3cc7-438a-4ee9-aa6d-bd371c266436"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score on X_train data set is 0.8244193955883782 \n",
      "AUC score on X_test data set is 0.7635679271708684 \n"
     ]
    }
   ],
   "source": [
    "y_pred=best_model.predict_proba(X_train)\n",
    "print(f\"AUC score on X_train data set is {roc_auc_score(y_train,y_pred[:,1])} \")\n",
    "y_pred=best_model.predict_proba(X_test)\n",
    "print(f'AUC score on X_test data set is {roc_auc_score(y_test,y_pred[:,1])} ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "id": "o3wTZZOoSnS1"
   },
   "outputs": [],
   "source": [
    "y_pred=best_model.predict_proba(X_train)\n",
    "fpr,tpr,thre=roc_curve(y_train,y_pred.T[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "id": "KTk5sbqDSoXG"
   },
   "outputs": [],
   "source": [
    "best_threshold=thre[np.argmax(tpr*(1-fpr))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "id": "u5uISOrdSpjN"
   },
   "outputs": [],
   "source": [
    "def best_thre(x,t):\n",
    "    return list(map(lambda x:1 if x>=t else 0,x.T[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 281
    },
    "id": "qaUz5MtQSrM-",
    "outputId": "511b44f3-970a-4aca-fc10-4834831a627e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred=best_model.predict_proba(X_train)\n",
    "y_pred=best_thre(y_pred,best_threshold)\n",
    "train_cm=pd.DataFrame(confusion_matrix(y_train,y_pred).T,index=['Predicted 0','Predicted 1'],columns=['actual 0','actual 1'])\n",
    "heatmap(train_cm,annot=True,fmt='g')\n",
    "plt.title('confusion matrix for train data(threshould={})'.format(round(best_threshold,2)))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 281
    },
    "id": "WW1UZr1XSsR2",
    "outputId": "43cbd690-4dbe-4a60-f00f-00328deebb48"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred=best_model.predict_proba(X_test)\n",
    "y_pred=best_thre(y_pred,best_threshold)\n",
    "train_cm=pd.DataFrame(confusion_matrix(y_test,y_pred).T,index=['Predicted 0','Predicted 1'],columns=['actual 0','actual 1'])\n",
    "heatmap(train_cm,annot=True,fmt='g')\n",
    "plt.title('confusion matrix for test data(threshould={})'.format(round(best_threshold,2)))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7lZVtXcES4jW"
   },
   "source": [
    "Using Decomposed U matrix we got AUC score of 0.763 using logistic regression model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "id": "G4bKhCTsStVG"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Copy of Recommendation_system_assignment_final.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}