lab_jupyter_content_clustering .ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p style=\"text-align:center\">\n",
    "    <a href=\"https://skills.network/?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMML321ENSkillsNetwork817-2022-01-01\" target=\"_blank\">\n",
    "    <img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/assets/logos/SN_web_lightmode.png\" width=\"200\" alt=\"Skills Network Logo\"  />\n",
    "    </a>\n",
    "</p>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# **Clustering based Course Recommender System**\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Estimated time needed: **90** minutes\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Previously, we have generated user profile vectors based on course ratings and genres. \n",
    "\n",
    "A user profile vector may look like a row vector in the following matrix, for example, we can see the Database column for user2 has a value 1 which means user2 is very interesting in courses related to the databases. With the user profile vectors generated, we can also easily compute the similarity among users based on their shared interests.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/userprofiles.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Furthermore, we could perform clustering algorithms such as K-means or DBSCAN to group users with similar learning interests. For example, in the below user clusters, we have user clusters whom have learned courses related to machine learning, cloud computing, databases, and web development, etc.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/userprofiles_clustering.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each user group, we can come up with a list of popular courses. For example, for the machine learning user cluster/learning group, we can count the most frequently enrolled courses, which are very likely to be the most popular and good machine learning courses because they are enrolled by many users who are interested in machine learning. \n",
    "\n",
    "If we know a user belongs to the machine learning group, we may recommend the most enrolled courses to them and it is very likely the user will be interested in them.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next in this lab, you will be implementing some clustering-based recommender system algorithms.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Objectives\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "After completing this lab you will be able to:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* Perform k-means clustering on the original user profile feature vectors\n",
    "* Apply PCA (Principle Component Analysis ) on user profile feature vectors to reduce dimensions\n",
    "* Perform k-means clustering on the PCA transformed main components\n",
    "* Generate course recommendations based on other group members' enrollment history\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "----\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare and setup lab environment\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First install and import required libraries:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: scikit-learn==1.0.2 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (1.0.2)\n",
      "Requirement already satisfied: joblib>=0.11 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from scikit-learn==1.0.2) (1.2.0)\n",
      "Requirement already satisfied: scipy>=1.1.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from scikit-learn==1.0.2) (1.7.3)\n",
      "Requirement already satisfied: numpy>=1.14.6 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from scikit-learn==1.0.2) (1.21.6)\n",
      "Requirement already satisfied: threadpoolctl>=2.0.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from scikit-learn==1.0.2) (3.1.0)\n",
      "Requirement already satisfied: seaborn==0.11.1 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (0.11.1)\n",
      "Requirement already satisfied: pandas>=0.23 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from seaborn==0.11.1) (1.3.5)\n",
      "Requirement already satisfied: matplotlib>=2.2 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from seaborn==0.11.1) (3.5.3)\n",
      "Requirement already satisfied: scipy>=1.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from seaborn==0.11.1) (1.7.3)\n",
      "Requirement already satisfied: numpy>=1.15 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from seaborn==0.11.1) (1.21.6)\n",
      "Requirement already satisfied: python-dateutil>=2.7 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (2.8.2)\n",
      "Requirement already satisfied: packaging>=20.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (21.3)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (0.11.0)\n",
      "Requirement already satisfied: pyparsing>=2.2.1 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (3.0.9)\n",
      "Requirement already satisfied: pillow>=6.2.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (8.1.0)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (1.4.4)\n",
      "Requirement already satisfied: fonttools>=4.22.0 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn==0.11.1) (4.38.0)\n",
      "Requirement already satisfied: pytz>=2017.3 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from pandas>=0.23->seaborn==0.11.1) (2022.6)\n",
      "Requirement already satisfied: typing-extensions in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from kiwisolver>=1.0.1->matplotlib>=2.2->seaborn==0.11.1) (4.4.0)\n",
      "Requirement already satisfied: six>=1.5 in /home/jupyterlab/conda/envs/python/lib/python3.7/site-packages (from python-dateutil>=2.7->matplotlib>=2.2->seaborn==0.11.1) (1.16.0)\n"
     ]
    }
   ],
   "source": [
    "!pip install scikit-learn==1.0.2\n",
    "!pip install seaborn==0.11.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.decomposition import PCA\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# also set a random state\n",
    "rs = 123"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load the user profile dataset\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's first load the original user profile feature vectors:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>user</th>\n",
       "      <th>Database</th>\n",
       "      <th>Python</th>\n",
       "      <th>CloudComputing</th>\n",
       "      <th>DataAnalysis</th>\n",
       "      <th>Containers</th>\n",
       "      <th>MachineLearning</th>\n",
       "      <th>ComputerVision</th>\n",
       "      <th>DataScience</th>\n",
       "      <th>BigData</th>\n",
       "      <th>Chatbot</th>\n",
       "      <th>R</th>\n",
       "      <th>BackendDev</th>\n",
       "      <th>FrontendDev</th>\n",
       "      <th>Blockchain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>52.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>43.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>33.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>40.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>24.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>14.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user  Database  Python  CloudComputing  DataAnalysis  Containers  \\\n",
       "0     2      52.0    14.0             6.0          43.0         3.0   \n",
       "1     4      40.0     2.0             4.0          28.0         0.0   \n",
       "2     5      24.0     8.0            18.0          24.0         0.0   \n",
       "3     7       2.0     0.0             0.0           2.0         0.0   \n",
       "4     8       6.0     0.0             0.0           4.0         0.0   \n",
       "\n",
       "   MachineLearning  ComputerVision  DataScience  BigData  Chatbot     R  \\\n",
       "0             33.0             0.0         29.0     41.0      2.0  18.0   \n",
       "1             14.0             0.0         20.0     24.0      0.0   6.0   \n",
       "2             30.0             0.0         22.0     14.0      2.0  14.0   \n",
       "3              0.0             0.0          0.0      2.0      0.0   0.0   \n",
       "4              0.0             0.0          0.0      6.0      0.0   2.0   \n",
       "\n",
       "   BackendDev  FrontendDev  Blockchain  \n",
       "0        34.0          9.0         6.0  \n",
       "1         6.0          0.0         2.0  \n",
       "2        26.0          4.0         6.0  \n",
       "3         0.0          0.0         0.0  \n",
       "4         0.0          0.0         0.0  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profile_url = \"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/datasets/user_profile.csv\"\n",
    "user_profile_df = pd.read_csv(user_profile_url)\n",
    "user_profile_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33901, 15)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profile_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we can then list the feature names, they are the user interested topics (course genres):\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Database',\n",
       " 'Python',\n",
       " 'CloudComputing',\n",
       " 'DataAnalysis',\n",
       " 'Containers',\n",
       " 'MachineLearning',\n",
       " 'ComputerVision',\n",
       " 'DataScience',\n",
       " 'BigData',\n",
       " 'Chatbot',\n",
       " 'R',\n",
       " 'BackendDev',\n",
       " 'FrontendDev',\n",
       " 'Blockchain']"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_names = list(user_profile_df.columns[1:])\n",
    "feature_names"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see from the user profile dataset, we have about 33K unique users with interests in areas like ``Database``, ``Python``, ``CloudComputing``, etc. Then, let's check the  summary statistics for each feature.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>user</th>\n",
       "      <th>Database</th>\n",
       "      <th>Python</th>\n",
       "      <th>CloudComputing</th>\n",
       "      <th>DataAnalysis</th>\n",
       "      <th>Containers</th>\n",
       "      <th>MachineLearning</th>\n",
       "      <th>ComputerVision</th>\n",
       "      <th>DataScience</th>\n",
       "      <th>BigData</th>\n",
       "      <th>Chatbot</th>\n",
       "      <th>R</th>\n",
       "      <th>BackendDev</th>\n",
       "      <th>FrontendDev</th>\n",
       "      <th>Blockchain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "      <td>33901.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.064064e+06</td>\n",
       "      <td>5.518569</td>\n",
       "      <td>3.493791</td>\n",
       "      <td>2.307100</td>\n",
       "      <td>3.624701</td>\n",
       "      <td>0.998938</td>\n",
       "      <td>3.048022</td>\n",
       "      <td>0.001770</td>\n",
       "      <td>5.087343</td>\n",
       "      <td>4.750450</td>\n",
       "      <td>0.462995</td>\n",
       "      <td>1.017433</td>\n",
       "      <td>2.269461</td>\n",
       "      <td>0.315802</td>\n",
       "      <td>0.986844</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>4.972578e+05</td>\n",
       "      <td>7.611941</td>\n",
       "      <td>4.227254</td>\n",
       "      <td>3.841858</td>\n",
       "      <td>4.760135</td>\n",
       "      <td>2.351764</td>\n",
       "      <td>4.624004</td>\n",
       "      <td>0.072846</td>\n",
       "      <td>5.230697</td>\n",
       "      <td>7.216228</td>\n",
       "      <td>1.055186</td>\n",
       "      <td>2.342532</td>\n",
       "      <td>4.310752</td>\n",
       "      <td>1.363462</td>\n",
       "      <td>1.950585</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>2.000000e+00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>6.813480e+05</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.043907e+06</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.451159e+06</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>2.103039e+06</td>\n",
       "      <td>63.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>38.000000</td>\n",
       "      <td>48.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>39.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>12.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               user      Database        Python  CloudComputing  DataAnalysis  \\\n",
       "count  3.390100e+04  33901.000000  33901.000000    33901.000000  33901.000000   \n",
       "mean   1.064064e+06      5.518569      3.493791        2.307100      3.624701   \n",
       "std    4.972578e+05      7.611941      4.227254        3.841858      4.760135   \n",
       "min    2.000000e+00      0.000000      0.000000        0.000000      0.000000   \n",
       "25%    6.813480e+05      0.000000      0.000000        0.000000      0.000000   \n",
       "50%    1.043907e+06      3.000000      3.000000        0.000000      3.000000   \n",
       "75%    1.451159e+06      9.000000      6.000000        3.000000      6.000000   \n",
       "max    2.103039e+06     63.000000     18.000000       38.000000     48.000000   \n",
       "\n",
       "         Containers  MachineLearning  ComputerVision   DataScience  \\\n",
       "count  33901.000000     33901.000000    33901.000000  33901.000000   \n",
       "mean       0.998938         3.048022        0.001770      5.087343   \n",
       "std        2.351764         4.624004        0.072846      5.230697   \n",
       "min        0.000000         0.000000        0.000000      0.000000   \n",
       "25%        0.000000         0.000000        0.000000      0.000000   \n",
       "50%        0.000000         0.000000        0.000000      3.000000   \n",
       "75%        0.000000         3.000000        0.000000      9.000000   \n",
       "max       15.000000        39.000000        3.000000     32.000000   \n",
       "\n",
       "            BigData       Chatbot             R    BackendDev   FrontendDev  \\\n",
       "count  33901.000000  33901.000000  33901.000000  33901.000000  33901.000000   \n",
       "mean       4.750450      0.462995      1.017433      2.269461      0.315802   \n",
       "std        7.216228      1.055186      2.342532      4.310752      1.363462   \n",
       "min        0.000000      0.000000      0.000000      0.000000      0.000000   \n",
       "25%        0.000000      0.000000      0.000000      0.000000      0.000000   \n",
       "50%        2.000000      0.000000      0.000000      0.000000      0.000000   \n",
       "75%        6.000000      0.000000      0.000000      3.000000      0.000000   \n",
       "max       54.000000      6.000000     18.000000     54.000000     15.000000   \n",
       "\n",
       "         Blockchain  \n",
       "count  33901.000000  \n",
       "mean       0.986844  \n",
       "std        1.950585  \n",
       "min        0.000000  \n",
       "25%        0.000000  \n",
       "50%        0.000000  \n",
       "75%        2.000000  \n",
       "max       12.000000  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profile_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The original user profile feature vector is not normalized, which may cause issues when we perform clustering and Principal component analysis (PCA), therefor we standardize the data.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean Database          -2.217793e-14\n",
      "Python             7.204451e-14\n",
      "CloudComputing    -3.443061e-14\n",
      "DataAnalysis      -2.543096e-15\n",
      "Containers         2.028543e-14\n",
      "MachineLearning   -1.199377e-14\n",
      "ComputerVision     3.136613e-15\n",
      "DataScience        5.723643e-15\n",
      "BigData           -4.646660e-15\n",
      "Chatbot            1.207204e-14\n",
      "R                 -5.597167e-15\n",
      "BackendDev        -7.880864e-15\n",
      "FrontendDev       -2.312337e-14\n",
      "Blockchain         4.286451e-14\n",
      "dtype: float64 and standard deviationDatabase           1.000015\n",
      "Python             1.000015\n",
      "CloudComputing     1.000015\n",
      "DataAnalysis       1.000015\n",
      "Containers         1.000015\n",
      "MachineLearning    1.000015\n",
      "ComputerVision     1.000015\n",
      "DataScience        1.000015\n",
      "BigData            1.000015\n",
      "Chatbot            1.000015\n",
      "R                  1.000015\n",
      "BackendDev         1.000015\n",
      "FrontendDev        1.000015\n",
      "Blockchain         1.000015\n",
      "dtype: float64 \n"
     ]
    }
   ],
   "source": [
    "# Use StandardScaler to make each feature with mean 0, standard deviation 1\n",
    "scaler = StandardScaler()\n",
    "user_profile_df[feature_names] = scaler.fit_transform(user_profile_df[feature_names])\n",
    "print(\"mean {} and standard deviation{} \".format(user_profile_df[feature_names].mean(),user_profile_df[feature_names].std()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>user</th>\n",
       "      <th>Database</th>\n",
       "      <th>Python</th>\n",
       "      <th>CloudComputing</th>\n",
       "      <th>DataAnalysis</th>\n",
       "      <th>Containers</th>\n",
       "      <th>MachineLearning</th>\n",
       "      <th>ComputerVision</th>\n",
       "      <th>DataScience</th>\n",
       "      <th>BigData</th>\n",
       "      <th>Chatbot</th>\n",
       "      <th>R</th>\n",
       "      <th>BackendDev</th>\n",
       "      <th>FrontendDev</th>\n",
       "      <th>Blockchain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "      <td>3.390100e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.064064e+06</td>\n",
       "      <td>-2.217793e-14</td>\n",
       "      <td>7.204451e-14</td>\n",
       "      <td>-3.443061e-14</td>\n",
       "      <td>-2.543096e-15</td>\n",
       "      <td>2.028543e-14</td>\n",
       "      <td>-1.199377e-14</td>\n",
       "      <td>3.136613e-15</td>\n",
       "      <td>5.723643e-15</td>\n",
       "      <td>-4.646660e-15</td>\n",
       "      <td>1.207204e-14</td>\n",
       "      <td>-5.597167e-15</td>\n",
       "      <td>-7.880864e-15</td>\n",
       "      <td>-2.312337e-14</td>\n",
       "      <td>4.286451e-14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>4.972578e+05</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "      <td>1.000015e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>2.000000e+00</td>\n",
       "      <td>-7.249991e-01</td>\n",
       "      <td>-8.265040e-01</td>\n",
       "      <td>-6.005256e-01</td>\n",
       "      <td>-7.614816e-01</td>\n",
       "      <td>-4.247674e-01</td>\n",
       "      <td>-6.591835e-01</td>\n",
       "      <td>-2.429612e-02</td>\n",
       "      <td>-9.726080e-01</td>\n",
       "      <td>-6.583107e-01</td>\n",
       "      <td>-4.387873e-01</td>\n",
       "      <td>-4.343370e-01</td>\n",
       "      <td>-5.264730e-01</td>\n",
       "      <td>-2.316212e-01</td>\n",
       "      <td>-5.059295e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>6.813480e+05</td>\n",
       "      <td>-7.249991e-01</td>\n",
       "      <td>-8.265040e-01</td>\n",
       "      <td>-6.005256e-01</td>\n",
       "      <td>-7.614816e-01</td>\n",
       "      <td>-4.247674e-01</td>\n",
       "      <td>-6.591835e-01</td>\n",
       "      <td>-2.429612e-02</td>\n",
       "      <td>-9.726080e-01</td>\n",
       "      <td>-6.583107e-01</td>\n",
       "      <td>-4.387873e-01</td>\n",
       "      <td>-4.343370e-01</td>\n",
       "      <td>-5.264730e-01</td>\n",
       "      <td>-2.316212e-01</td>\n",
       "      <td>-5.059295e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.043907e+06</td>\n",
       "      <td>-3.308757e-01</td>\n",
       "      <td>-1.168130e-01</td>\n",
       "      <td>-6.005256e-01</td>\n",
       "      <td>-1.312380e-01</td>\n",
       "      <td>-4.247674e-01</td>\n",
       "      <td>-6.591835e-01</td>\n",
       "      <td>-2.429612e-02</td>\n",
       "      <td>-3.990622e-01</td>\n",
       "      <td>-3.811535e-01</td>\n",
       "      <td>-4.387873e-01</td>\n",
       "      <td>-4.343370e-01</td>\n",
       "      <td>-5.264730e-01</td>\n",
       "      <td>-2.316212e-01</td>\n",
       "      <td>-5.059295e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.451159e+06</td>\n",
       "      <td>4.573712e-01</td>\n",
       "      <td>5.928781e-01</td>\n",
       "      <td>1.803581e-01</td>\n",
       "      <td>4.990056e-01</td>\n",
       "      <td>-4.247674e-01</td>\n",
       "      <td>-1.038556e-02</td>\n",
       "      <td>-2.429612e-02</td>\n",
       "      <td>7.480294e-01</td>\n",
       "      <td>1.731609e-01</td>\n",
       "      <td>-4.387873e-01</td>\n",
       "      <td>-4.343370e-01</td>\n",
       "      <td>1.694715e-01</td>\n",
       "      <td>-2.316212e-01</td>\n",
       "      <td>5.194189e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>2.103039e+06</td>\n",
       "      <td>7.551593e+00</td>\n",
       "      <td>3.431642e+00</td>\n",
       "      <td>9.290667e+00</td>\n",
       "      <td>9.322416e+00</td>\n",
       "      <td>5.953517e+00</td>\n",
       "      <td>7.775190e+00</td>\n",
       "      <td>4.115884e+01</td>\n",
       "      <td>5.145214e+00</td>\n",
       "      <td>6.824934e+00</td>\n",
       "      <td>5.247500e+00</td>\n",
       "      <td>7.249771e+00</td>\n",
       "      <td>1.200053e+01</td>\n",
       "      <td>1.076995e+01</td>\n",
       "      <td>5.646161e+00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               user      Database        Python  CloudComputing  DataAnalysis  \\\n",
       "count  3.390100e+04  3.390100e+04  3.390100e+04    3.390100e+04  3.390100e+04   \n",
       "mean   1.064064e+06 -2.217793e-14  7.204451e-14   -3.443061e-14 -2.543096e-15   \n",
       "std    4.972578e+05  1.000015e+00  1.000015e+00    1.000015e+00  1.000015e+00   \n",
       "min    2.000000e+00 -7.249991e-01 -8.265040e-01   -6.005256e-01 -7.614816e-01   \n",
       "25%    6.813480e+05 -7.249991e-01 -8.265040e-01   -6.005256e-01 -7.614816e-01   \n",
       "50%    1.043907e+06 -3.308757e-01 -1.168130e-01   -6.005256e-01 -1.312380e-01   \n",
       "75%    1.451159e+06  4.573712e-01  5.928781e-01    1.803581e-01  4.990056e-01   \n",
       "max    2.103039e+06  7.551593e+00  3.431642e+00    9.290667e+00  9.322416e+00   \n",
       "\n",
       "         Containers  MachineLearning  ComputerVision   DataScience  \\\n",
       "count  3.390100e+04     3.390100e+04    3.390100e+04  3.390100e+04   \n",
       "mean   2.028543e-14    -1.199377e-14    3.136613e-15  5.723643e-15   \n",
       "std    1.000015e+00     1.000015e+00    1.000015e+00  1.000015e+00   \n",
       "min   -4.247674e-01    -6.591835e-01   -2.429612e-02 -9.726080e-01   \n",
       "25%   -4.247674e-01    -6.591835e-01   -2.429612e-02 -9.726080e-01   \n",
       "50%   -4.247674e-01    -6.591835e-01   -2.429612e-02 -3.990622e-01   \n",
       "75%   -4.247674e-01    -1.038556e-02   -2.429612e-02  7.480294e-01   \n",
       "max    5.953517e+00     7.775190e+00    4.115884e+01  5.145214e+00   \n",
       "\n",
       "            BigData       Chatbot             R    BackendDev   FrontendDev  \\\n",
       "count  3.390100e+04  3.390100e+04  3.390100e+04  3.390100e+04  3.390100e+04   \n",
       "mean  -4.646660e-15  1.207204e-14 -5.597167e-15 -7.880864e-15 -2.312337e-14   \n",
       "std    1.000015e+00  1.000015e+00  1.000015e+00  1.000015e+00  1.000015e+00   \n",
       "min   -6.583107e-01 -4.387873e-01 -4.343370e-01 -5.264730e-01 -2.316212e-01   \n",
       "25%   -6.583107e-01 -4.387873e-01 -4.343370e-01 -5.264730e-01 -2.316212e-01   \n",
       "50%   -3.811535e-01 -4.387873e-01 -4.343370e-01 -5.264730e-01 -2.316212e-01   \n",
       "75%    1.731609e-01 -4.387873e-01 -4.343370e-01  1.694715e-01 -2.316212e-01   \n",
       "max    6.824934e+00  5.247500e+00  7.249771e+00  1.200053e+01  1.076995e+01   \n",
       "\n",
       "         Blockchain  \n",
       "count  3.390100e+04  \n",
       "mean   4.286451e-14  \n",
       "std    1.000015e+00  \n",
       "min   -5.059295e-01  \n",
       "25%   -5.059295e-01  \n",
       "50%   -5.059295e-01  \n",
       "75%    5.194189e-01  \n",
       "max    5.646161e+00  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_profile_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The normalized user profile features are: \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>Database</th>\n",
       "      <th>Python</th>\n",
       "      <th>CloudComputing</th>\n",
       "      <th>DataAnalysis</th>\n",
       "      <th>Containers</th>\n",
       "      <th>MachineLearning</th>\n",
       "      <th>ComputerVision</th>\n",
       "      <th>DataScience</th>\n",
       "      <th>BigData</th>\n",
       "      <th>Chatbot</th>\n",
       "      <th>R</th>\n",
       "      <th>BackendDev</th>\n",
       "      <th>FrontendDev</th>\n",
       "      <th>Blockchain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6.106474</td>\n",
       "      <td>2.485388</td>\n",
       "      <td>0.961242</td>\n",
       "      <td>8.272010</td>\n",
       "      <td>0.850889</td>\n",
       "      <td>6.477594</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>4.571668</td>\n",
       "      <td>5.023412</td>\n",
       "      <td>1.456642</td>\n",
       "      <td>7.249771</td>\n",
       "      <td>7.360899</td>\n",
       "      <td>6.369323</td>\n",
       "      <td>2.570116</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.529980</td>\n",
       "      <td>-0.353377</td>\n",
       "      <td>0.440653</td>\n",
       "      <td>5.120792</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>2.368540</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>2.851031</td>\n",
       "      <td>2.667576</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>2.127032</td>\n",
       "      <td>0.865416</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>0.519419</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.427988</td>\n",
       "      <td>1.066006</td>\n",
       "      <td>4.084776</td>\n",
       "      <td>4.280467</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>5.828796</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>3.233395</td>\n",
       "      <td>1.281790</td>\n",
       "      <td>1.456642</td>\n",
       "      <td>5.542191</td>\n",
       "      <td>5.505047</td>\n",
       "      <td>2.702132</td>\n",
       "      <td>2.570116</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.462250</td>\n",
       "      <td>-0.826504</td>\n",
       "      <td>-0.600526</td>\n",
       "      <td>-0.341319</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.659184</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>-0.972608</td>\n",
       "      <td>-0.381153</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>-0.434337</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.063248</td>\n",
       "      <td>-0.826504</td>\n",
       "      <td>-0.600526</td>\n",
       "      <td>0.078843</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.659184</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>-0.972608</td>\n",
       "      <td>0.173161</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>0.419453</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33896</th>\n",
       "      <td>-0.330876</td>\n",
       "      <td>-0.116813</td>\n",
       "      <td>0.180358</td>\n",
       "      <td>0.499006</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.659184</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>0.174484</td>\n",
       "      <td>-0.658311</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>2.127032</td>\n",
       "      <td>0.865416</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>1.032093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33897</th>\n",
       "      <td>-0.724999</td>\n",
       "      <td>-0.826504</td>\n",
       "      <td>-0.079936</td>\n",
       "      <td>-0.761482</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.659184</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>-0.972608</td>\n",
       "      <td>-0.658311</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>-0.434337</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33898</th>\n",
       "      <td>-0.330876</td>\n",
       "      <td>0.592878</td>\n",
       "      <td>0.961242</td>\n",
       "      <td>-0.761482</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>2.368540</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>0.748029</td>\n",
       "      <td>-0.658311</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>0.846348</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33899</th>\n",
       "      <td>-0.724999</td>\n",
       "      <td>-0.826504</td>\n",
       "      <td>-0.600526</td>\n",
       "      <td>-0.761482</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.226652</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>-0.972608</td>\n",
       "      <td>-0.658311</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>-0.434337</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33900</th>\n",
       "      <td>-0.724999</td>\n",
       "      <td>-0.826504</td>\n",
       "      <td>-0.600526</td>\n",
       "      <td>-0.761482</td>\n",
       "      <td>-0.424767</td>\n",
       "      <td>-0.226652</td>\n",
       "      <td>-0.024296</td>\n",
       "      <td>-0.972608</td>\n",
       "      <td>-0.658311</td>\n",
       "      <td>-0.438787</td>\n",
       "      <td>-0.434337</td>\n",
       "      <td>-0.526473</td>\n",
       "      <td>-0.231621</td>\n",
       "      <td>-0.505930</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>33901 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Database    Python  CloudComputing  DataAnalysis  Containers  \\\n",
       "0      6.106474  2.485388        0.961242      8.272010    0.850889   \n",
       "1      4.529980 -0.353377        0.440653      5.120792   -0.424767   \n",
       "2      2.427988  1.066006        4.084776      4.280467   -0.424767   \n",
       "3     -0.462250 -0.826504       -0.600526     -0.341319   -0.424767   \n",
       "4      0.063248 -0.826504       -0.600526      0.078843   -0.424767   \n",
       "...         ...       ...             ...           ...         ...   \n",
       "33896 -0.330876 -0.116813        0.180358      0.499006   -0.424767   \n",
       "33897 -0.724999 -0.826504       -0.079936     -0.761482   -0.424767   \n",
       "33898 -0.330876  0.592878        0.961242     -0.761482   -0.424767   \n",
       "33899 -0.724999 -0.826504       -0.600526     -0.761482   -0.424767   \n",
       "33900 -0.724999 -0.826504       -0.600526     -0.761482   -0.424767   \n",
       "\n",
       "       MachineLearning  ComputerVision  DataScience   BigData   Chatbot  \\\n",
       "0             6.477594       -0.024296     4.571668  5.023412  1.456642   \n",
       "1             2.368540       -0.024296     2.851031  2.667576 -0.438787   \n",
       "2             5.828796       -0.024296     3.233395  1.281790  1.456642   \n",
       "3            -0.659184       -0.024296    -0.972608 -0.381153 -0.438787   \n",
       "4            -0.659184       -0.024296    -0.972608  0.173161 -0.438787   \n",
       "...                ...             ...          ...       ...       ...   \n",
       "33896        -0.659184       -0.024296     0.174484 -0.658311 -0.438787   \n",
       "33897        -0.659184       -0.024296    -0.972608 -0.658311 -0.438787   \n",
       "33898         2.368540       -0.024296     0.748029 -0.658311 -0.438787   \n",
       "33899        -0.226652       -0.024296    -0.972608 -0.658311 -0.438787   \n",
       "33900        -0.226652       -0.024296    -0.972608 -0.658311 -0.438787   \n",
       "\n",
       "              R  BackendDev  FrontendDev  Blockchain  \n",
       "0      7.249771    7.360899     6.369323    2.570116  \n",
       "1      2.127032    0.865416    -0.231621    0.519419  \n",
       "2      5.542191    5.505047     2.702132    2.570116  \n",
       "3     -0.434337   -0.526473    -0.231621   -0.505930  \n",
       "4      0.419453   -0.526473    -0.231621   -0.505930  \n",
       "...         ...         ...          ...         ...  \n",
       "33896  2.127032    0.865416    -0.231621    1.032093  \n",
       "33897 -0.434337   -0.526473    -0.231621   -0.505930  \n",
       "33898  0.846348   -0.526473    -0.231621   -0.505930  \n",
       "33899 -0.434337   -0.526473    -0.231621   -0.505930  \n",
       "33900 -0.434337   -0.526473    -0.231621   -0.505930  \n",
       "\n",
       "[33901 rows x 14 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features = user_profile_df.loc[:, user_profile_df.columns != 'user']\n",
    "features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we can also save the user ids for later recommendation tasks:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>user</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33896</th>\n",
       "      <td>2102054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33897</th>\n",
       "      <td>2102356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33898</th>\n",
       "      <td>2102680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33899</th>\n",
       "      <td>2102983</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33900</th>\n",
       "      <td>2103039</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>33901 rows × 1 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          user\n",
       "0            2\n",
       "1            4\n",
       "2            5\n",
       "3            7\n",
       "4            8\n",
       "...        ...\n",
       "33896  2102054\n",
       "33897  2102356\n",
       "33898  2102680\n",
       "33899  2102983\n",
       "33900  2103039\n",
       "\n",
       "[33901 rows x 1 columns]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_ids = user_profile_df.loc[:, user_profile_df.columns == 'user']\n",
    "user_ids"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TASK: Perform K-means clustering algorithm on the user profile feature vectors\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With the user profile dataset ready, you need to use the `KMeans` class provided by scikit-learn library to perform clustering on the user profile feature vectors. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For `KMeans` algorithm, one important hyperparameter is the number of clusters `n_cluster`, and a good way to find the optimized `n_cluster` is using to grid search a list of candidates and find the one with the best or optimized clustering evaluation metrics such as minimal `sum of squared distance`:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: grid search the optimized n_cluster for KMeans() model_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# WRITE YOUR CODE HERE\n",
    "from sklearn.cluster import KMeans\n",
    "\n",
    "km_list = list()\n",
    "# Find an optimized number of neighors k from a candidate list such as list_k = list(range(1, 30))\n",
    "for clust in range(1,30):\n",
    "    km = KMeans(n_clusters=clust, random_state=42)\n",
    "    km = km.fit(features)\n",
    "    \n",
    "    km_list.append(pd.Series({'clusters': clust, \n",
    "                              'inertia': km.inertia_,\n",
    "                              'model': km}))\n",
    "\n",
    "\n",
    "# plotting the k-means \n",
    "\n",
    "plot_data = (pd.concat(km_list, axis=1)\n",
    "             .T\n",
    "             [['clusters','inertia']]\n",
    "             .set_index('clusters'))\n",
    "\n",
    "ax = plot_data.plot(marker='o',ls='-')\n",
    "ax.set_xticks(range(0,21,2))\n",
    "ax.set_xlim(0,21)\n",
    "ax.set(xlabel='Cluster', ylabel='Inertia');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details>\n",
    "    <summary>Click here for Hints</summary>\n",
    "    \n",
    "Create a list that will hold the the sum of square distances for each fitted model. For each k in `n_clusters` make a model by calling `KMeans(n_clusters=k, random_state=rs).fit(features)` ans append `model.inertia_` to the list. Plot square distances against the k values.  \n",
    "\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you plot the grid search process, you may get a elbow plot like the following:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/clusters_elbow.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the elbow plot, you should visualy identify the point where the metric starting to be flatten, which indicates the optimized number of clusters.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once you have identified the best number of clusters, you can apply `KMeans()` again to generate cluster label for all users.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "cluster_labels = [None] * len(user_ids)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: Apply KMeans() on the features with optimized n_cluster parameter after model fitting, you can find output cluster labels in `model.labels_` attribute_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "## WRITE YOUR CODE HERE\n",
    "km =KMeans(n_clusters=18, random_state = 42)\n",
    "km.fit(features)\n",
    "## ...\n",
    "cluster_labels = km.labels_\n",
    "## ...\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33901,)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "km.labels_.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details>\n",
    "    <summary>Click here for Hints</summary>\n",
    "    \n",
    "Create  a model by calling `KMeans(n_clusters=k, random_state=rs).fit(features)`. Save the labels by accessing `model.labels`.\n",
    "\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The cluster labels you generated is a list of integers indicating cluster indices. You may use the following utility method to combine the cluster labels and user ids to a dataframe, so that you know which cluster a user belongs:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def combine_cluster_labels(user_ids, labels):\n",
    "    labels_df = pd.DataFrame(labels)\n",
    "    cluster_df = pd.merge(user_ids, labels_df, left_index=True, right_index=True)\n",
    "    cluster_df.columns = ['user', 'cluster']\n",
    "    return cluster_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>user</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33896</th>\n",
       "      <td>2102054</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33897</th>\n",
       "      <td>2102356</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33898</th>\n",
       "      <td>2102680</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33899</th>\n",
       "      <td>2102983</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33900</th>\n",
       "      <td>2103039</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>33901 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          user  cluster\n",
       "0            2       15\n",
       "1            4       11\n",
       "2            5       15\n",
       "3            7        7\n",
       "4            8        7\n",
       "...        ...      ...\n",
       "33896  2102054        9\n",
       "33897  2102356        7\n",
       "33898  2102680       17\n",
       "33899  2102983        7\n",
       "33900  2103039        7\n",
       "\n",
       "[33901 rows x 2 columns]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cluster_df = combine_cluster_labels(user_ids, cluster_labels)\n",
    "cluster_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Your clustering results may look like the following screenshot:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/cluster_labels.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, each user finds its own cluster or we can say we have created many clusters of learning communities. Learners within each community share very similar learning interests.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TASK: Apply PCA on user profile feature vectors to reduce dimensions\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the previous step, we applied `KMeans` on the original user profile feature vectors which have 14 original features (the course genres).\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = user_profile_df.loc[:, user_profile_df.columns != 'user']\n",
    "user_ids = user_profile_df.loc[:, user_profile_df.columns == 'user']\n",
    "feature_names = list(user_profile_df.columns[1:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 14 features for each user profile.\n"
     ]
    }
   ],
   "source": [
    "print(f\"There are {len(feature_names)} features for each user profile.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we plot a covariance matrix of the user profile feature vectors with 14 features, we can observe that some features are actually correlated:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1100x900 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_theme(style=\"white\")\n",
    "\n",
    "# Compute the correlation matrix\n",
    "corr = features.cov()\n",
    "\n",
    "# Generate a mask for the upper triangle\n",
    "mask = np.triu(np.ones_like(corr, dtype=bool))\n",
    "\n",
    "# Set up the matplotlib figure\n",
    "f, ax = plt.subplots(figsize=(11, 9))\n",
    "\n",
    "# Generate a custom diverging colormap\n",
    "cmap = sns.diverging_palette(230, 20, as_cmap=True)\n",
    "\n",
    "# Draw the heatmap with the mask and correct aspect ratio\n",
    "sns.heatmap(corr, mask=mask, cmap=cmap, vmax=.3, center=0,\n",
    "            square=True, linewidths=.5, cbar_kws={\"shrink\": .5})\n",
    "\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For example, the feature `MachineLearning` and the feature `DataScience` are correlated. Such covariances among features may indicate that we can apply PCA to find its main components (eigenvectors with max eigenvalues on the covariance matrix). \n",
    "\n",
    "If we only keep the independent main components, then we can reduce the dimensions of our user profile feature vectors.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's apply the `PCA()` provided by  `scikit-learn` to find the main components in user profile feature vectors and see if we can reduce its dimensions by only keeping the main components.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that when calling the  `PCA()` class, there is also an import argument called `n_components` which indicates how many components you want to keep in the PCA result. One way to find an optimized `n_components` is to do a grid search on a list of argument candidates (such as `range(1, 15)`) and calculate the ratio of the accumulated variance for each candidate. \n",
    "\n",
    "If the accumulated variances ratio of a candidate `n_components` is larger than a threshold, e.g., 90%, then we can say the transformed `n_components` could explain about 90% of variances of the original data variance and can be considered as an optimized components size.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: Find the optimized `n_components` for PCA_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>model</th>\n",
       "      <th>var</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>n</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>PCA(n_components=1)</td>\n",
       "      <td>0.287568</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>PCA(n_components=2)</td>\n",
       "      <td>0.463093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>PCA(n_components=3)</td>\n",
       "      <td>0.576134</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>PCA(n_components=4)</td>\n",
       "      <td>0.648747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>PCA(n_components=5)</td>\n",
       "      <td>0.719488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>PCA(n_components=6)</td>\n",
       "      <td>0.78821</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>PCA(n_components=7)</td>\n",
       "      <td>0.842956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>PCA(n_components=8)</td>\n",
       "      <td>0.894445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>PCA(n_components=9)</td>\n",
       "      <td>0.92727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>PCA(n_components=10)</td>\n",
       "      <td>0.954466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>PCA(n_components=11)</td>\n",
       "      <td>0.973357</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>PCA(n_components=12)</td>\n",
       "      <td>0.987031</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>PCA(n_components=13)</td>\n",
       "      <td>0.997916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>PCA(n_components=14)</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   model       var\n",
       "n                                 \n",
       "1    PCA(n_components=1)  0.287568\n",
       "2    PCA(n_components=2)  0.463093\n",
       "3    PCA(n_components=3)  0.576134\n",
       "4    PCA(n_components=4)  0.648747\n",
       "5    PCA(n_components=5)  0.719488\n",
       "6    PCA(n_components=6)   0.78821\n",
       "7    PCA(n_components=7)  0.842956\n",
       "8    PCA(n_components=8)  0.894445\n",
       "9    PCA(n_components=9)   0.92727\n",
       "10  PCA(n_components=10)  0.954466\n",
       "11  PCA(n_components=11)  0.973357\n",
       "12  PCA(n_components=12)  0.987031\n",
       "13  PCA(n_components=13)  0.997916\n",
       "14  PCA(n_components=14)       1.0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# WRITE YOUR CODE HERE\n",
    "\n",
    "# - For a list of candidate `n_components` arguments such as 1 to 14, find out the minimal `n` that can explain accumulated 90% variances of previous data\n",
    "# - In the fitted PCA() model, you can find explained_variance_ratio_ and use the sum() function to add them to get the accumulated variance ratio\n",
    "\n",
    "from sklearn.decomposition import PCA\n",
    "# Fit a range of PCA models\n",
    "pca_list = list()\n",
    "    \n",
    "for n in range(1, 15):\n",
    "    \n",
    "    # Create and fit the model\n",
    "    PCAmod = PCA(n_components=n)\n",
    "    PCAmod.fit(features)\n",
    "    \n",
    "    # Store the model and variance\n",
    "    pca_list.append(pd.Series({'n': n, \n",
    "                               'model': PCAmod,\n",
    "                               'var': PCAmod.explained_variance_ratio_.sum()}))\n",
    "    \n",
    "pca_df = pd.concat(pca_list, axis=1).T.set_index('n')\n",
    "pca_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.set_context('talk')\n",
    "ax = pca_df['var'].plot(kind='bar')\n",
    "plt.hlines(y=0.9,xmin=0.5,xmax=14.5,color='red')\n",
    "ax.set(xlabel='components',\n",
    "       ylabel='Accumulated Variance Ratios',\n",
    "       title='Explained Variance vs Components');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details>\n",
    "    <summary>Click here for Hints</summary>\n",
    "    \n",
    "* For each `n_components` from 1 to 14 you can call `PCA=PCA(n_components=component)`, then you can simply fit it by callsing `pca.fit_transform(features)` where `features = user_profile_df.loc[:, user_profile_df.columns != 'user']`. \n",
    "* Then you can find `accumulated_variance_ratios` by applying `sum()` to `pca.explained_variance_ratio_`. \n",
    "* Then find the smallest n_components value for which `accumulated_variance_ratios >= 0.9` and return it.\n",
    "</details>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you visualize your hyperparameter searching process, you may get a trend line like the following:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/pca.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once you found the optimized `n_component` argument value, you can apply PCA on the user profile feature vectors and reduce the  14 features into  `n_component` features.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: Perform PCA to transform original user profile features_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[17.77249439,  0.20068141,  1.73060927, ...,  0.681042  ,\n",
       "         2.31261304,  0.86827151],\n",
       "       [ 7.14519863, -2.84748121,  2.3586356 , ...,  0.54976933,\n",
       "         0.46903346,  0.03340487],\n",
       "       [11.36326985,  1.8736192 , -1.52207738, ...,  1.67758234,\n",
       "         2.9376687 ,  2.09763858],\n",
       "       ...,\n",
       "       [ 0.62594328, -0.54716673, -1.69282423, ..., -0.05509971,\n",
       "         0.58209105,  1.70319253],\n",
       "       [-2.03683187, -0.15353383,  0.16285226, ...,  0.07249638,\n",
       "         0.11374975,  0.62289958],\n",
       "       [-2.03683187, -0.15353383,  0.16285226, ...,  0.07249638,\n",
       "         0.11374975,  0.62289958]])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# WRITE YOUR CODE HERE\n",
    "\n",
    "# - For a list of candidate `n_components` arguments such as 1 to 14, find out the minimal `n` that can explain accumulated 90% variances of previous data\n",
    "# - In the fitted PCA() model, you can find explained_variance_ratio_ and use the sum() function to add them to get the accumulated variance ratio\n",
    "# - Merge the user ids and transformed features into a new dataframe\n",
    "PCAmod = PCA(n_components=9) \n",
    "feature_pca = PCAmod.fit_transform(features)\n",
    "\n",
    "feature_pca\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "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>user</th>\n",
       "      <th>pc0</th>\n",
       "      <th>pc1</th>\n",
       "      <th>pc2</th>\n",
       "      <th>pc3</th>\n",
       "      <th>pc4</th>\n",
       "      <th>pc5</th>\n",
       "      <th>pc6</th>\n",
       "      <th>pc7</th>\n",
       "      <th>pc8</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>17.772494</td>\n",
       "      <td>0.200681</td>\n",
       "      <td>1.730609</td>\n",
       "      <td>2.567359</td>\n",
       "      <td>-3.825814</td>\n",
       "      <td>2.707154</td>\n",
       "      <td>0.681042</td>\n",
       "      <td>2.312613</td>\n",
       "      <td>0.868272</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>7.145199</td>\n",
       "      <td>-2.847481</td>\n",
       "      <td>2.358636</td>\n",
       "      <td>-0.576654</td>\n",
       "      <td>0.398803</td>\n",
       "      <td>-0.134533</td>\n",
       "      <td>0.549769</td>\n",
       "      <td>0.469033</td>\n",
       "      <td>0.033405</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>11.363270</td>\n",
       "      <td>1.873619</td>\n",
       "      <td>-1.522077</td>\n",
       "      <td>1.076144</td>\n",
       "      <td>-1.711688</td>\n",
       "      <td>0.883212</td>\n",
       "      <td>1.677582</td>\n",
       "      <td>2.937669</td>\n",
       "      <td>2.097639</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>-1.834033</td>\n",
       "      <td>-0.277462</td>\n",
       "      <td>0.564905</td>\n",
       "      <td>0.053470</td>\n",
       "      <td>-0.064440</td>\n",
       "      <td>0.165757</td>\n",
       "      <td>0.030956</td>\n",
       "      <td>0.039519</td>\n",
       "      <td>0.210887</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>-1.049125</td>\n",
       "      <td>-0.684767</td>\n",
       "      <td>1.072765</td>\n",
       "      <td>0.006371</td>\n",
       "      <td>-0.005695</td>\n",
       "      <td>0.118686</td>\n",
       "      <td>0.118559</td>\n",
       "      <td>0.559292</td>\n",
       "      <td>0.186379</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user        pc0       pc1       pc2       pc3       pc4       pc5  \\\n",
       "0     2  17.772494  0.200681  1.730609  2.567359 -3.825814  2.707154   \n",
       "1     4   7.145199 -2.847481  2.358636 -0.576654  0.398803 -0.134533   \n",
       "2     5  11.363270  1.873619 -1.522077  1.076144 -1.711688  0.883212   \n",
       "3     7  -1.834033 -0.277462  0.564905  0.053470 -0.064440  0.165757   \n",
       "4     8  -1.049125 -0.684767  1.072765  0.006371 -0.005695  0.118686   \n",
       "\n",
       "        pc6       pc7       pc8  \n",
       "0  0.681042  2.312613  0.868272  \n",
       "1  0.549769  0.469033  0.033405  \n",
       "2  1.677582  2.937669  2.097639  \n",
       "3  0.030956  0.039519  0.210887  \n",
       "4  0.118559  0.559292  0.186379  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_pca_df = pd.DataFrame(feature_pca,)\n",
    "pca_df = pd.merge(user_ids, feature_pca_df, left_index=True, right_index=True)\n",
    "pca_df.columns = ['user', 'pc0', 'pc1', 'pc2', 'pc3', 'pc4', 'pc5', 'pc6', 'pc7', 'pc8']\n",
    "pca_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details>\n",
    "    <summary>Click here for Hints</summary>\n",
    "    \n",
    "* Call PCA class as `pca = PCA(n_components=n_components)` \n",
    "* Fit PCA model using predefined `features` variable as only parameter\n",
    "* Get the components by calling `pca.fit_transform(features)` \n",
    "* Create a `pd.DataFrame(data=components)` and use `pd.merge` to merge it with `user_ids` don't forget to specify `left_index=True, right_index=True` in `merge` function parameters.\n",
    "    \n",
    "</details> \n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Your PCA transformed dataframe may look like the following:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/pca_res.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TASK: Perform k-means clustering on the PCA transformed feature vectors\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, you have the PCA  components of the original profile vectors. You can perform k-means on them again:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: Perform K-means on the PCA transformed features_\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>user</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33896</th>\n",
       "      <td>2102054</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33897</th>\n",
       "      <td>2102356</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33898</th>\n",
       "      <td>2102680</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33899</th>\n",
       "      <td>2102983</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33900</th>\n",
       "      <td>2103039</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>33901 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          user  cluster\n",
       "0            2       11\n",
       "1            4       11\n",
       "2            5       11\n",
       "3            7       11\n",
       "4            8       11\n",
       "...        ...      ...\n",
       "33896  2102054        3\n",
       "33897  2102356        3\n",
       "33898  2102680        3\n",
       "33899  2102983        3\n",
       "33900  2103039        3\n",
       "\n",
       "[33901 rows x 2 columns]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## - Obtain the cluster label lists from model.labels_ attribute\n",
    "km =KMeans(n_clusters=18, random_state = 42)\n",
    "km.fit(pca_df)\n",
    "## ...\n",
    "cluster_labels = km.labels_\n",
    "\n",
    "\n",
    "\n",
    "## - Assign each user a cluster label by combining user ids and cluster labels\n",
    "cluster_df = combine_cluster_labels(user_ids, cluster_labels)\n",
    "cluster_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Your clustering results should have the same format as the k-means on the original dataset:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/cluster_labels.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Great, now all users find their learning interest groups, either based on their original or the PCA transformed user profile features. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When a user is in a group or a community, it is very likely that the user will be interested in the courses enrolled by other members within the same group.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TASK: Generate course recommendations based on the popular courses in the same cluster\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Intuition of clustering-based course recommendation is very simple and can be illustrated via the following example:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Suppose a user has joined a machine learning group (via clustering algorithm). In the group, he/she finds that the top-3 courses enrolled by all other group members are `Machine Learning for Everyone`, `Machine Learning with Python`, `Machine Learning with Scikit-learn`. Since the user has already completed the `Machine Learning for Everyone` earlier, he/she decides to trust the group members' choices and enroll in other two unselected courses `Machine Learning with Python` and `Machine Learning with Scikit-learn`.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In summary, the clustering-based recommender system first groups all users based on their profiles, and maintains a popular courses list for each group. \n",
    "\n",
    "For any group member who needs course recommendations, the algorithm recommends the unselected courses from the popular course lists.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, suppose we have a set of test users, and we want to recommend new courses to them using a clustering-based recommender system:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>user</th>\n",
       "      <th>item</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1502801</td>\n",
       "      <td>RP0105EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1609720</td>\n",
       "      <td>CNSC02EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1347188</td>\n",
       "      <td>CO0301EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>755067</td>\n",
       "      <td>ML0103EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>538595</td>\n",
       "      <td>BD0115EN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      user      item\n",
       "0  1502801  RP0105EN\n",
       "1  1609720  CNSC02EN\n",
       "2  1347188  CO0301EN\n",
       "3   755067  ML0103EN\n",
       "4   538595  BD0115EN"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_user_url = \"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/datasets/rs_content_test.csv\"\n",
    "test_users_df = pd.read_csv(test_user_url)[['user', 'item']]\n",
    "test_users_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The test users dataset has only two columns, the user id and the enrolled course id. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each user, let's find its cluster label using the k-means results you have performed in previous steps, assuming it is named `cluster_df`.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can assign the cluster label to all test users via merging the clustering labels (`cluster_df`:):\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "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>user</th>\n",
       "      <th>item</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1502801</td>\n",
       "      <td>RP0105EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1502801</td>\n",
       "      <td>BD0131EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1502801</td>\n",
       "      <td>BD0212EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1502801</td>\n",
       "      <td>BD0115EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1502801</td>\n",
       "      <td>BD0211EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9397</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0121EN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9398</th>\n",
       "      <td>630511</td>\n",
       "      <td>SC0101EN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9399</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9400</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0115EN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9401</th>\n",
       "      <td>630511</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9402 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         user      item  cluster\n",
       "0     1502801  RP0105EN        1\n",
       "1     1502801  BD0131EN        1\n",
       "2     1502801  BD0212EN        1\n",
       "3     1502801  BD0115EN        1\n",
       "4     1502801  BD0211EN        1\n",
       "...       ...       ...      ...\n",
       "9397   630511  BD0121EN        0\n",
       "9398   630511  SC0101EN        0\n",
       "9399   630511  BD0111EN        0\n",
       "9400   630511  BD0115EN        0\n",
       "9401   630511  PY0101EN        0\n",
       "\n",
       "[9402 rows x 3 columns]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_users_labelled = pd.merge(test_users_df, cluster_df, left_on='user', right_on='user')\n",
    "test_users_labelled"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The merged the test dataset may look like the following:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBM-ML321EN-SkillsNetwork/labs/module_3/images/test_users_label.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From the above dataframe, we know each user's enrolled courses and its cluster index.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we use a `groupby`  and `sum` aggregation, we can get the enrollments count for each course in each group, like the following code snippet:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\ncourses_cluster = test_users_labelled[['item', 'cluster']]\\ncourses_cluster['count'] = [1] * len(courses_cluster)\\ncourses_cluster.groupby(['cluster','item']).agg(enrollments = ('count','sum')).reset_index()\\n\\n\""
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    "courses_cluster = test_users_labelled[['item', 'cluster']]\n",
    "courses_cluster['count'] = [1] * len(courses_cluster)\n",
    "courses_cluster.groupby(['cluster','item']).agg(enrollments = ('count','sum')).reset_index()\n",
    "\n",
    "'''\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "_TODO: For each test user, try to recommend any unseen courses based on the popular courses in his/her cluster. You may use an enrollment count threshold (such as larger than 10) to determine if it is a popular course in the cluster_ \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "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>cluster</th>\n",
       "      <th>item</th>\n",
       "      <th>enrollments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>LB0101ENv1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7</td>\n",
       "      <td>BD0131EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>7</td>\n",
       "      <td>BD0135EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>BD0141EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7</td>\n",
       "      <td>BD0212EN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1271</th>\n",
       "      <td>0</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1272</th>\n",
       "      <td>14</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1273</th>\n",
       "      <td>8</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1274</th>\n",
       "      <td>8</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>60</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1275</th>\n",
       "      <td>8</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>60</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1276 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      cluster        item  enrollments\n",
       "0           7  LB0101ENv1            1\n",
       "1           7    BD0131EN            1\n",
       "2           7    BD0135EN            1\n",
       "3           7    BD0141EN            1\n",
       "4           7    BD0212EN            1\n",
       "...       ...         ...          ...\n",
       "1271        0    BD0101EN           51\n",
       "1272       14    BD0101EN           51\n",
       "1273        8    BD0101EN           57\n",
       "1274        8    DS0101EN           60\n",
       "1275        8    PY0101EN           60\n",
       "\n",
       "[1276 rows x 3 columns]"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "courses_cluster = test_users_labelled[['item', 'cluster']]\n",
    "courses_cluster['count'] = [1] * len(courses_cluster)\n",
    "courses_cluster = courses_cluster.groupby(['cluster','item']).agg(enrollments = ('count','sum')).sort_values(by='enrollments').reset_index()\n",
    "courses_cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>index</th>\n",
       "      <th>cluster</th>\n",
       "      <th>item</th>\n",
       "      <th>enrollments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1229</td>\n",
       "      <td>15</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1230</td>\n",
       "      <td>1</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1231</td>\n",
       "      <td>2</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1232</td>\n",
       "      <td>6</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1233</td>\n",
       "      <td>17</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1234</td>\n",
       "      <td>6</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1235</td>\n",
       "      <td>5</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1236</td>\n",
       "      <td>8</td>\n",
       "      <td>CB0103EN</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1237</td>\n",
       "      <td>14</td>\n",
       "      <td>BD0211EN</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1238</td>\n",
       "      <td>2</td>\n",
       "      <td>CO0101EN</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1239</td>\n",
       "      <td>13</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1240</td>\n",
       "      <td>12</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1241</td>\n",
       "      <td>5</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1242</td>\n",
       "      <td>2</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1243</td>\n",
       "      <td>14</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>1244</td>\n",
       "      <td>8</td>\n",
       "      <td>DA0101EN</td>\n",
       "      <td>36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>1245</td>\n",
       "      <td>8</td>\n",
       "      <td>DS0105EN</td>\n",
       "      <td>36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1246</td>\n",
       "      <td>0</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1247</td>\n",
       "      <td>8</td>\n",
       "      <td>ML0115EN</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>1248</td>\n",
       "      <td>17</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>1249</td>\n",
       "      <td>8</td>\n",
       "      <td>BC0101EN</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>1250</td>\n",
       "      <td>15</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>1251</td>\n",
       "      <td>15</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>1252</td>\n",
       "      <td>16</td>\n",
       "      <td>ML0101ENv3</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>1253</td>\n",
       "      <td>6</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>1254</td>\n",
       "      <td>5</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>1255</td>\n",
       "      <td>16</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>1256</td>\n",
       "      <td>8</td>\n",
       "      <td>DS0103EN</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>1257</td>\n",
       "      <td>5</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>1258</td>\n",
       "      <td>10</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>1259</td>\n",
       "      <td>0</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>1260</td>\n",
       "      <td>13</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>1261</td>\n",
       "      <td>14</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>1262</td>\n",
       "      <td>2</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>1263</td>\n",
       "      <td>2</td>\n",
       "      <td>BC0101EN</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>1264</td>\n",
       "      <td>14</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>1265</td>\n",
       "      <td>13</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>43</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>1266</td>\n",
       "      <td>0</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>1267</td>\n",
       "      <td>8</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>1268</td>\n",
       "      <td>16</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>1269</td>\n",
       "      <td>17</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>1270</td>\n",
       "      <td>13</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>1271</td>\n",
       "      <td>0</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>1272</td>\n",
       "      <td>14</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>1273</td>\n",
       "      <td>8</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>1274</td>\n",
       "      <td>8</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>60</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>1275</td>\n",
       "      <td>8</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>60</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    index  cluster        item  enrollments\n",
       "0    1229       15    DS0101EN           31\n",
       "1    1230        1    PY0101EN           32\n",
       "2    1231        2    DS0101EN           32\n",
       "3    1232        6    DS0101EN           32\n",
       "4    1233       17    BD0101EN           32\n",
       "5    1234        6    BD0101EN           32\n",
       "6    1235        5    PY0101EN           32\n",
       "7    1236        8    CB0103EN           33\n",
       "8    1237       14    BD0211EN           33\n",
       "9    1238        2    CO0101EN           33\n",
       "10   1239       13    BD0111EN           33\n",
       "11   1240       12    PY0101EN           34\n",
       "12   1241        5    BD0111EN           34\n",
       "13   1242        2    BD0101EN           35\n",
       "14   1243       14    PY0101EN           35\n",
       "15   1244        8    DA0101EN           36\n",
       "16   1245        8    DS0105EN           36\n",
       "17   1246        0    PY0101EN           37\n",
       "18   1247        8    ML0115EN           37\n",
       "19   1248       17    DS0101EN           37\n",
       "20   1249        8    BC0101EN           38\n",
       "21   1250       15    BD0111EN           38\n",
       "22   1251       15    BD0101EN           38\n",
       "23   1252       16  ML0101ENv3           38\n",
       "24   1253        6    PY0101EN           38\n",
       "25   1254        5    DS0101EN           38\n",
       "26   1255       16    DS0101EN           40\n",
       "27   1256        8    DS0103EN           40\n",
       "28   1257        5    BD0101EN           40\n",
       "29   1258       10    PY0101EN           41\n",
       "30   1259        0    BD0111EN           41\n",
       "31   1260       13    DS0101EN           42\n",
       "32   1261       14    DS0101EN           42\n",
       "33   1262        2    PY0101EN           43\n",
       "34   1263        2    BC0101EN           43\n",
       "35   1264       14    BD0111EN           43\n",
       "36   1265       13    PY0101EN           43\n",
       "37   1266        0    DS0101EN           44\n",
       "38   1267        8    BD0111EN           45\n",
       "39   1268       16    PY0101EN           45\n",
       "40   1269       17    PY0101EN           46\n",
       "41   1270       13    BD0101EN           47\n",
       "42   1271        0    BD0101EN           51\n",
       "43   1272       14    BD0101EN           51\n",
       "44   1273        8    BD0101EN           57\n",
       "45   1274        8    DS0101EN           60\n",
       "46   1275        8    PY0101EN           60"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "enrollment_threshold = 30\n",
    "Popular_courses_cluster = courses_cluster[courses_cluster['enrollments']>enrollment_threshold].reset_index()\n",
    "Popular_courses_cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0       DS0101EN\n",
       "1       PY0101EN\n",
       "2       DS0101EN\n",
       "3       DS0101EN\n",
       "4       BD0101EN\n",
       "5       BD0101EN\n",
       "6       PY0101EN\n",
       "7       CB0103EN\n",
       "8       BD0211EN\n",
       "9       CO0101EN\n",
       "10      BD0111EN\n",
       "11      PY0101EN\n",
       "12      BD0111EN\n",
       "13      BD0101EN\n",
       "14      PY0101EN\n",
       "15      DA0101EN\n",
       "16      DS0105EN\n",
       "17      PY0101EN\n",
       "18      ML0115EN\n",
       "19      DS0101EN\n",
       "20      BC0101EN\n",
       "21      BD0111EN\n",
       "22      BD0101EN\n",
       "23    ML0101ENv3\n",
       "24      PY0101EN\n",
       "25      DS0101EN\n",
       "26      DS0101EN\n",
       "27      DS0103EN\n",
       "28      BD0101EN\n",
       "29      PY0101EN\n",
       "30      BD0111EN\n",
       "31      DS0101EN\n",
       "32      DS0101EN\n",
       "33      PY0101EN\n",
       "34      BC0101EN\n",
       "35      BD0111EN\n",
       "36      PY0101EN\n",
       "37      DS0101EN\n",
       "38      BD0111EN\n",
       "39      PY0101EN\n",
       "40      PY0101EN\n",
       "41      BD0101EN\n",
       "42      BD0101EN\n",
       "43      BD0101EN\n",
       "44      BD0101EN\n",
       "45      DS0101EN\n",
       "46      PY0101EN\n",
       "Name: item, dtype: object"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Popular_courses = Popular_courses_cluster['item']\n",
    "Popular_courses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "## WRITE YOUR CODE HERE\n",
    "\n",
    "\n",
    "user=[]\n",
    "courses=[]\n",
    "\n",
    "\n",
    "courses_cluster = test_users_labelled[['item', 'cluster']]\n",
    "courses_cluster['count'] = [1] * len(courses_cluster)\n",
    "courses_cluster = courses_cluster.groupby(['cluster','item']).agg(enrollments = ('count','sum')).sort_values(by='enrollments').reset_index()\n",
    "enrollment_threshold = 30  # setting a threshold which can be changed\n",
    "Popular_cluster = courses_cluster[courses_cluster['enrollments']>enrollment_threshold].reset_index()\n",
    "#Popular_courses = Popular_cluster['item']\n",
    "\n",
    "    \n",
    "## - For each user, first finds its test_users_labelled['user'] == user_id cluster label\n",
    "for user_d in test_users_labelled['user']:\n",
    "    user_d = test_users_labelled[test_users_labelled['user'] == user_d]\n",
    "    enrolled_course_cluster = user_d['cluster'].iloc[0]  #the cluster of enrolled course\n",
    "    \n",
    "    ## - First get all courses belonging to the same cluster and figure out what are the popular ones (such as course enrollments beyond a threshold like 100)\n",
    "    all_courses_in_cluster = courses_cluster[courses_cluster['cluster'] == enrolled_course_cluster]['item'].to_list()\n",
    "    all_courses_in_cluster = set(all_courses_in_cluster)\n",
    "    \n",
    "    ## - Get the user's current enrolled courses\n",
    "    enrolled_courses = user_d['item'].to_list()\n",
    "    enrolled_courses = set(enrolled_courses)\n",
    "    \n",
    "    ## - Check if there are any courses on the popular course list which are new/unseen to the user. \n",
    "    \n",
    "    \n",
    "    unknown_courses = all_courses_in_cluster.difference(enrolled_courses)\n",
    "    unknown_course_df = test_users_labelled[test_users_labelled['item'].isin(unknown_courses)]\n",
    "    unknown_course_ids = unknown_course_df['item'].values\n",
    "    \n",
    "    \n",
    "    \n",
    "    Popular_courses_cluster = Popular_cluster[Popular_cluster['cluster']==enrolled_course_cluster].reset_index()\n",
    "    \n",
    "    recommended_courses_df = Popular_courses_cluster[Popular_courses_cluster['item'].isin(set(unknown_course_ids))]\n",
    "    recommended_courses_df = recommended_courses_df[recommended_courses_df['cluster']== enrolled_course_cluster]['item']\n",
    "    recommended_courses_ids = recommended_courses_df.values\n",
    "   \n",
    "    \n",
    "    \n",
    "    user_id = user_d['user'].iloc[0]\n",
    "    \n",
    "    ## If yes, make those unseen and popular courses as recommendation results for the user\n",
    "    for i in range(0, len(recommended_courses_ids)):\n",
    "        user.append(user_id)\n",
    "        courses.append(recommended_courses_ids[i])\n",
    "    \n",
    "\n",
    "    \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [],
   "source": [
    "rec = {}\n",
    "rec['USER']= user\n",
    "rec['COURSES']= courses\n",
    "rec_df = pd.DataFrame(rec, columns=['USER', 'COURSES'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "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>USER</th>\n",
       "      <th>COURSES</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1502801</td>\n",
       "      <td>PY0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1502801</td>\n",
       "      <td>PY0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1502801</td>\n",
       "      <td>PY0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1502801</td>\n",
       "      <td>PY0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1502801</td>\n",
       "      <td>PY0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13123</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13124</th>\n",
       "      <td>630511</td>\n",
       "      <td>DS0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13125</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13126</th>\n",
       "      <td>630511</td>\n",
       "      <td>DS0101EN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13127</th>\n",
       "      <td>630511</td>\n",
       "      <td>BD0101EN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>13128 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          USER   COURSES\n",
       "0      1502801  PY0101EN\n",
       "1      1502801  PY0101EN\n",
       "2      1502801  PY0101EN\n",
       "3      1502801  PY0101EN\n",
       "4      1502801  PY0101EN\n",
       "...        ...       ...\n",
       "13123   630511  BD0101EN\n",
       "13124   630511  DS0101EN\n",
       "13125   630511  BD0101EN\n",
       "13126   630511  DS0101EN\n",
       "13127   630511  BD0101EN\n",
       "\n",
       "[13128 rows x 2 columns]"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rec_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "USER\n",
       "899349     170\n",
       "970895     120\n",
       "919445     112\n",
       "927126     108\n",
       "945602     104\n",
       "          ... \n",
       "1669829      5\n",
       "1190880      5\n",
       "1560987      5\n",
       "1180992      5\n",
       "464965       5\n",
       "Name: COURSES, Length: 629, dtype: int64"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user = rec_df.groupby('USER')['COURSES'].size().sort_values(ascending=False) # number of courses recommended to each user\n",
    "user"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20.87122416534181"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user.mean() #average course recommended per user"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "COURSES\n",
       "PY0101EN      2524\n",
       "DS0101EN      2304\n",
       "BD0101EN      2013\n",
       "BD0111EN      1535\n",
       "BC0101EN       979\n",
       "CB0103EN       644\n",
       "ML0115EN       589\n",
       "DA0101EN       560\n",
       "DS0105EN       551\n",
       "DS0103EN       519\n",
       "CO0101EN       373\n",
       "BD0211EN       294\n",
       "ML0101ENv3     243\n",
       "Name: USER, dtype: int64"
      ]
     },
     "execution_count": 179,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cour = rec_df.groupby('COURSES')['USER'].size().sort_values(ascending=False) # number of users a course is recommended to\n",
    "cour"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1009.8461538461538"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cour.mean() #average number of users a course is recommended to"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "COURSES\n",
       "PY0101EN    2524\n",
       "DS0101EN    2304\n",
       "BD0101EN    2013\n",
       "BD0111EN    1535\n",
       "BC0101EN     979\n",
       "CB0103EN     644\n",
       "ML0115EN     589\n",
       "DA0101EN     560\n",
       "DS0105EN     551\n",
       "DS0103EN     519\n",
       "Name: USER, dtype: int64"
      ]
     },
     "execution_count": 181,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cour_top = cour.head(10) #Top ten courses\n",
    "cour_top"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "<details>\n",
    "    <summary>Click here for Hints</summary>\n",
    "    \n",
    "* First of all, create a `user_subset` of  `test_users_labelled` where `test_users_labelled['user'] == user_id`. \n",
    "* Get the enrolled courses by  simply accessing `['item']` column of `user_subset`\n",
    "* Find its cluster label by accessing `['cluster']` column of `user_subset`. You can just use the first one (`.iloc[0]`) since every value in the column is the same for an individual user.\n",
    "* You can find all courses in the same cluster, by simply accessing `['item']` column of a subset of `test_users_labelled` where `test_users_labelled[test_users_labelled['cluster'] == cluster_id]`\n",
    "* You can find new/unseen courses to the user by taking a set difference between courses in cluster and enrolled_courses by using `.difference` method (dont forget to convert 2 lists in to sets before calling the method). \n",
    "* Use `courses_cluster` data set to find popularity of new/unseen courses and make unseen and popular courses as recommendation results for the user\n",
    "    \n",
    "</details> \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "630511"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['RP0105EN', 'BD0131EN', 'BD0212EN', ..., 'LB0109ENv1', 'DS0101EN',\n",
       "       'CO0101EN'], dtype=object)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unknown_course_ids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7353"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(unknown_course_ids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "tags": []
   },
   "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>level_0</th>\n",
       "      <th>index</th>\n",
       "      <th>cluster</th>\n",
       "      <th>item</th>\n",
       "      <th>enrollments</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>17</td>\n",
       "      <td>1246</td>\n",
       "      <td>0</td>\n",
       "      <td>PY0101EN</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>30</td>\n",
       "      <td>1259</td>\n",
       "      <td>0</td>\n",
       "      <td>BD0111EN</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>37</td>\n",
       "      <td>1266</td>\n",
       "      <td>0</td>\n",
       "      <td>DS0101EN</td>\n",
       "      <td>44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>42</td>\n",
       "      <td>1271</td>\n",
       "      <td>0</td>\n",
       "      <td>BD0101EN</td>\n",
       "      <td>51</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   level_0  index  cluster      item  enrollments\n",
       "0       17   1246        0  PY0101EN           37\n",
       "1       30   1259        0  BD0111EN           41\n",
       "2       37   1266        0  DS0101EN           44\n",
       "3       42   1271        0  BD0101EN           51"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Popular_courses_cluster"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With the recommendation results, you also need to write some analytic code to answer the following two questions:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "user_recommendations[courses_count]=sum(axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- On average, how many new/unseen courses have been recommended to each user?\n",
    "- What are the most frequently recommended courses? Return the top-10 commonly recommended courses across all users.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- User1: ['course1', 'course2']\n",
    "- User2: ['course3', 'course4']\n",
    "- User3: ['course3', 'course4', 'course5']\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "For example, suppose we have only 3 test users, each user receives the following recommendations:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then, the average recommended courses per user is $(2 + 2 + 3) / 3 = 2.33$. The top-2 recommended courses are: `course3`: 2 times, and `course4`: 2 times.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that the answers will depend on how you compute the popular courses for each cluster. A lower threshold yields more recommended courses but with smaller confidence so that some test users may receive very long course recommendation lists and feel overwhelmed. \n",
    "\n",
    "Ideally, we should limit the maximum course recommendations for each user to be less than 20 courses per user.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Explore other clustering algorithms\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you have learned in previous unsupervised learning course, there are many other clustering algorithms such as `DBSCAN` and `Hierarchical Clustering`. You are encouraged to try them on the user profile feature vectors and compare the results with K-means.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Congratulations! In this lab, you have applied clustering algorithms to group users with similar interests and also tried PCA to reduce the dimensions of user feature vectors.\n",
    "\n",
    "Furthermore, with each user finding its learning interest group, you have also implemented clustering-based course recommender system to make recommendations based on his/her group members' popular courses choices.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Authors\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[Yan Luo](https://www.linkedin.com/in/yan-luo-96288783/?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMML321ENSkillsNetwork817-2022-01-01)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Other Contributors\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Change Log\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "|Date (YYYY-MM-DD)|Version|Changed By|Change Description|\n",
    "|-|-|-|-|\n",
    "|2021-10-25|1.0|Yan|Created the initial version|\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Copyright © 2021 IBM Corporation. All rights reserved.\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python",
   "language": "python",
   "name": "conda-env-python-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}