06_group_segmentation.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"anaconda-cloud":{},"colab":{"name":"06_group_segmentation.ipynb","provenance":[],"collapsed_sections":[]},"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.8.8"}},"cells":[{"cell_type":"markdown","metadata":{"id":"XfvUGzgYlHES"},"source":["# Author's Notes\n","\n","May 2021 Update: We now generate tab-delimited files (TSVs) for each clustering algo. You can load the first three dimensions along with the labels in [Tensorboard Projector](https://projector.tensorflow.org) to visualize the group segmentation in 3-D space.\n","\n","From [Hands-on Unsupervised Learning using Python](https://amzn.to/2tVABNx) - Chapter 6 - Group Segmentation"]},{"cell_type":"markdown","metadata":{"id":"mFLRgf38laYc"},"source":["# Google Colab Setup\n","**Only for users on Google Colab**"]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"oFpjRFSSlfuu","executionInfo":{"status":"ok","timestamp":1620243368314,"user_tz":240,"elapsed":1340,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"d156255c-a93a-448f-cb66-86898855adb3"},"source":["# Define functions to connect to Google and change directories\n","def connectDrive():\n","    from google.colab import drive\n","    drive.mount('/content/drive', force_remount=True)\n","\n","def changeDirectory(path):\n","    import os\n","    original_path = os.getcwd()\n","    os.chdir(path)\n","    new_path = os.getcwd()\n","    print(\"Original path: \",original_path)\n","    print(\"New path: \",new_path)\n","\n","# Connect to Google Drive\n","connectDrive()\n","\n","# Change path\n","changeDirectory(\"/content/drive/My Drive/github/handson-unsupervised-learning/\")"],"execution_count":1,"outputs":[{"output_type":"stream","text":["Mounted at /content/drive\n","Original path:  /content\n","New path:  /content/drive/My Drive/github/handson-unsupervised-learning\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"eYZ9MHgHmPYl","executionInfo":{"status":"ok","timestamp":1620243374219,"user_tz":240,"elapsed":7233,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"8f0b9488-0d49-43db-cb13-03bb26df5420"},"source":["# Install additional libraries\n","!pip install fastcluster==1.1.26\n","!pip install hdbscan==0.8.27"],"execution_count":2,"outputs":[{"output_type":"stream","text":["Requirement already satisfied: fastcluster==1.1.26 in /usr/local/lib/python3.7/dist-packages (1.1.26)\n","Requirement already satisfied: numpy>=1.9 in /usr/local/lib/python3.7/dist-packages (from fastcluster==1.1.26) (1.19.5)\n","Requirement already satisfied: hdbscan==0.8.27 in /usr/local/lib/python3.7/dist-packages (0.8.27)\n","Requirement already satisfied: six in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (1.15.0)\n","Requirement already satisfied: joblib>=1.0 in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (1.0.1)\n","Requirement already satisfied: numpy>=1.16 in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (1.19.5)\n","Requirement already satisfied: scipy>=1.0 in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (1.4.1)\n","Requirement already satisfied: cython>=0.27 in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (0.29.22)\n","Requirement already satisfied: scikit-learn>=0.20 in /usr/local/lib/python3.7/dist-packages (from hdbscan==0.8.27) (0.22.2.post1)\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"lr3tDheV7kLT"},"source":["# Import Libraries"]},{"cell_type":"code","metadata":{"id":"q__4Yz54ze5u","executionInfo":{"status":"ok","timestamp":1620243374851,"user_tz":240,"elapsed":7855,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Import libraries\n","'''Main'''\n","import numpy as np\n","import pandas as pd\n","import os, time, re\n","import pickle, gzip\n","\n","'''Data Viz'''\n","import matplotlib.pyplot as plt\n","import matplotlib as mpl\n","import seaborn as sns\n","color = sns.color_palette()\n","%matplotlib inline\n","\n","'''Data Prep and Model Evaluation'''\n","from sklearn import preprocessing as pp\n","from sklearn import impute\n","from sklearn.model_selection import train_test_split \n","from sklearn.metrics import precision_recall_curve, average_precision_score\n","from sklearn.metrics import roc_curve, auc, roc_auc_score\n","\n","'''Algorithms'''\n","from sklearn.decomposition import PCA\n","from sklearn.cluster import KMeans\n","import fastcluster\n","from scipy.cluster.hierarchy import dendrogram, cophenet, fcluster\n","from scipy.spatial.distance import pdist"],"execution_count":3,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"4L-vERlz71z_"},"source":["# Load & Prepare Data"]},{"cell_type":"code","metadata":{"id":"olW8GFXvze5z","executionInfo":{"status":"ok","timestamp":1620243376050,"user_tz":240,"elapsed":9048,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Load the data\n","current_path = os.getcwd()\n","file = os.path.sep.join(['', 'datasets', 'lending_club_data', 'LoanStats3a.csv'])\n","data = pd.read_csv(current_path + file, low_memory=False)"],"execution_count":4,"outputs":[]},{"cell_type":"code","metadata":{"id":"GoFPxME2ze50","executionInfo":{"status":"ok","timestamp":1620243376051,"user_tz":240,"elapsed":9047,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Select columns to keep\n","columnsToKeep = ['loan_amnt','funded_amnt','funded_amnt_inv','term', \\\n","                 'int_rate','installment','grade','sub_grade', \\\n","                 'emp_length','home_ownership','annual_inc', \\\n","                 'verification_status','pymnt_plan','purpose', \\\n","                 'addr_state','dti','delinq_2yrs','earliest_cr_line', \\\n","                 'mths_since_last_delinq','mths_since_last_record', \\\n","                 'open_acc','pub_rec','revol_bal','revol_util', \\\n","                 'total_acc','initial_list_status','out_prncp', \\\n","                 'out_prncp_inv','total_pymnt','total_pymnt_inv', \\\n","                 'total_rec_prncp','total_rec_int','total_rec_late_fee', \\\n","                 'recoveries','collection_recovery_fee','last_pymnt_d', \\\n","                 'last_pymnt_amnt']\n","\n","data = data.loc[:,columnsToKeep]"],"execution_count":5,"outputs":[]},{"cell_type":"code","metadata":{"id":"D-VkWTYeze50","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243376051,"user_tz":240,"elapsed":9044,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"a662a66a-a4e5-4975-9b72-dc291a3529c9"},"source":["# View shape\n","data.shape"],"execution_count":6,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(42542, 37)"]},"metadata":{"tags":[]},"execution_count":6}]},{"cell_type":"code","metadata":{"id":"iKjQX3vjze51","colab":{"base_uri":"https://localhost:8080/","height":316},"executionInfo":{"status":"ok","timestamp":1620243376193,"user_tz":240,"elapsed":9176,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"56256cd7-07c6-45d2-8308-daa370d12882"},"source":["# Preview data\n","data.head()"],"execution_count":7,"outputs":[{"output_type":"execute_result","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>loan_amnt</th>\n","      <th>funded_amnt</th>\n","      <th>funded_amnt_inv</th>\n","      <th>term</th>\n","      <th>int_rate</th>\n","      <th>installment</th>\n","      <th>grade</th>\n","      <th>sub_grade</th>\n","      <th>emp_length</th>\n","      <th>home_ownership</th>\n","      <th>annual_inc</th>\n","      <th>verification_status</th>\n","      <th>pymnt_plan</th>\n","      <th>purpose</th>\n","      <th>addr_state</th>\n","      <th>dti</th>\n","      <th>delinq_2yrs</th>\n","      <th>earliest_cr_line</th>\n","      <th>mths_since_last_delinq</th>\n","      <th>mths_since_last_record</th>\n","      <th>open_acc</th>\n","      <th>pub_rec</th>\n","      <th>revol_bal</th>\n","      <th>revol_util</th>\n","      <th>total_acc</th>\n","      <th>initial_list_status</th>\n","      <th>out_prncp</th>\n","      <th>out_prncp_inv</th>\n","      <th>total_pymnt</th>\n","      <th>total_pymnt_inv</th>\n","      <th>total_rec_prncp</th>\n","      <th>total_rec_int</th>\n","      <th>total_rec_late_fee</th>\n","      <th>recoveries</th>\n","      <th>collection_recovery_fee</th>\n","      <th>last_pymnt_d</th>\n","      <th>last_pymnt_amnt</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>5000.0</td>\n","      <td>5000.0</td>\n","      <td>4975.0</td>\n","      <td>36 months</td>\n","      <td>10.65%</td>\n","      <td>162.87</td>\n","      <td>B</td>\n","      <td>B2</td>\n","      <td>10+ years</td>\n","      <td>RENT</td>\n","      <td>24000.0</td>\n","      <td>Verified</td>\n","      <td>n</td>\n","      <td>credit_card</td>\n","      <td>AZ</td>\n","      <td>27.65</td>\n","      <td>0.0</td>\n","      <td>Jan-85</td>\n","      <td>NaN</td>\n","      <td>NaN</td>\n","      <td>3.0</td>\n","      <td>0.0</td>\n","      <td>13648.0</td>\n","      <td>83.70%</td>\n","      <td>9.0</td>\n","      <td>f</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>5863.155187</td>\n","      <td>5833.84</td>\n","      <td>5000.00</td>\n","      <td>863.16</td>\n","      <td>0.00</td>\n","      <td>0.0</td>\n","      <td>0.00</td>\n","      <td>Jan-15</td>\n","      <td>171.62</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>2500.0</td>\n","      <td>2500.0</td>\n","      <td>2500.0</td>\n","      <td>60 months</td>\n","      <td>15.27%</td>\n","      <td>59.83</td>\n","      <td>C</td>\n","      <td>C4</td>\n","      <td>&lt; 1 year</td>\n","      <td>RENT</td>\n","      <td>30000.0</td>\n","      <td>Source Verified</td>\n","      <td>n</td>\n","      <td>car</td>\n","      <td>GA</td>\n","      <td>1.00</td>\n","      <td>0.0</td>\n","      <td>Apr-99</td>\n","      <td>NaN</td>\n","      <td>NaN</td>\n","      <td>3.0</td>\n","      <td>0.0</td>\n","      <td>1687.0</td>\n","      <td>9.40%</td>\n","      <td>4.0</td>\n","      <td>f</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>1014.530000</td>\n","      <td>1014.53</td>\n","      <td>456.46</td>\n","      <td>435.17</td>\n","      <td>0.00</td>\n","      <td>122.9</td>\n","      <td>1.11</td>\n","      <td>Apr-13</td>\n","      <td>119.66</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>2400.0</td>\n","      <td>2400.0</td>\n","      <td>2400.0</td>\n","      <td>36 months</td>\n","      <td>15.96%</td>\n","      <td>84.33</td>\n","      <td>C</td>\n","      <td>C5</td>\n","      <td>10+ years</td>\n","      <td>RENT</td>\n","      <td>12252.0</td>\n","      <td>Not Verified</td>\n","      <td>n</td>\n","      <td>small_business</td>\n","      <td>IL</td>\n","      <td>8.72</td>\n","      <td>0.0</td>\n","      <td>Nov-01</td>\n","      <td>NaN</td>\n","      <td>NaN</td>\n","      <td>2.0</td>\n","      <td>0.0</td>\n","      <td>2956.0</td>\n","      <td>98.50%</td>\n","      <td>10.0</td>\n","      <td>f</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>3005.666844</td>\n","      <td>3005.67</td>\n","      <td>2400.00</td>\n","      <td>605.67</td>\n","      <td>0.00</td>\n","      <td>0.0</td>\n","      <td>0.00</td>\n","      <td>Jun-14</td>\n","      <td>649.91</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>10000.0</td>\n","      <td>10000.0</td>\n","      <td>10000.0</td>\n","      <td>36 months</td>\n","      <td>13.49%</td>\n","      <td>339.31</td>\n","      <td>C</td>\n","      <td>C1</td>\n","      <td>10+ years</td>\n","      <td>RENT</td>\n","      <td>49200.0</td>\n","      <td>Source Verified</td>\n","      <td>n</td>\n","      <td>other</td>\n","      <td>CA</td>\n","      <td>20.00</td>\n","      <td>0.0</td>\n","      <td>Feb-96</td>\n","      <td>35.0</td>\n","      <td>NaN</td>\n","      <td>10.0</td>\n","      <td>0.0</td>\n","      <td>5598.0</td>\n","      <td>21%</td>\n","      <td>37.0</td>\n","      <td>f</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>12231.890000</td>\n","      <td>12231.89</td>\n","      <td>10000.00</td>\n","      <td>2214.92</td>\n","      <td>16.97</td>\n","      <td>0.0</td>\n","      <td>0.00</td>\n","      <td>Jan-15</td>\n","      <td>357.48</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>3000.0</td>\n","      <td>3000.0</td>\n","      <td>3000.0</td>\n","      <td>60 months</td>\n","      <td>12.69%</td>\n","      <td>67.79</td>\n","      <td>B</td>\n","      <td>B5</td>\n","      <td>1 year</td>\n","      <td>RENT</td>\n","      <td>80000.0</td>\n","      <td>Source Verified</td>\n","      <td>n</td>\n","      <td>other</td>\n","      <td>OR</td>\n","      <td>17.94</td>\n","      <td>0.0</td>\n","      <td>Jan-96</td>\n","      <td>38.0</td>\n","      <td>NaN</td>\n","      <td>15.0</td>\n","      <td>0.0</td>\n","      <td>27783.0</td>\n","      <td>53.90%</td>\n","      <td>38.0</td>\n","      <td>f</td>\n","      <td>0.0</td>\n","      <td>0.0</td>\n","      <td>4066.908161</td>\n","      <td>4066.91</td>\n","      <td>3000.00</td>\n","      <td>1066.91</td>\n","      <td>0.00</td>\n","      <td>0.0</td>\n","      <td>0.00</td>\n","      <td>Jan-17</td>\n","      <td>67.30</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["   loan_amnt  funded_amnt  ...  last_pymnt_d last_pymnt_amnt\n","0     5000.0       5000.0  ...        Jan-15          171.62\n","1     2500.0       2500.0  ...        Apr-13          119.66\n","2     2400.0       2400.0  ...        Jun-14          649.91\n","3    10000.0      10000.0  ...        Jan-15          357.48\n","4     3000.0       3000.0  ...        Jan-17           67.30\n","\n","[5 rows x 37 columns]"]},"metadata":{"tags":[]},"execution_count":7}]},{"cell_type":"code","metadata":{"id":"U1Gw2o42ze51","executionInfo":{"status":"ok","timestamp":1620243376709,"user_tz":240,"elapsed":9682,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Transform features from string to numeric\n","for i in [\"term\",\"int_rate\",\"emp_length\",\"revol_util\"]:\n","    data.loc[:,i] = \\\n","        data.loc[:,i].apply(lambda x: re.sub(\"[^0-9]\", \"\", str(x)))\n","    data.loc[:,i] = pd.to_numeric(data.loc[:,i])"],"execution_count":8,"outputs":[]},{"cell_type":"code","metadata":{"id":"_7LbVdq3ze51","executionInfo":{"status":"ok","timestamp":1620243376711,"user_tz":240,"elapsed":9682,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Determine which features are numerical\n","numericalFeats = [x for x in data.columns if data[x].dtype != 'object']"],"execution_count":9,"outputs":[]},{"cell_type":"code","metadata":{"id":"hDJSu02Eze52","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243376711,"user_tz":240,"elapsed":9680,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"b5f1d65a-edc4-41dc-9922-b5759526a5a2"},"source":["# Display NaNs by feature\n","nanCounter = np.isnan(data.loc[:,numericalFeats]).sum()\n","nanCounter"],"execution_count":10,"outputs":[{"output_type":"execute_result","data":{"text/plain":["loan_amnt                      7\n","funded_amnt                    7\n","funded_amnt_inv                7\n","term                           7\n","int_rate                       7\n","installment                    7\n","emp_length                  1119\n","annual_inc                    11\n","dti                            7\n","delinq_2yrs                   36\n","mths_since_last_delinq     26933\n","mths_since_last_record     38891\n","open_acc                      36\n","pub_rec                       36\n","revol_bal                      7\n","revol_util                    97\n","total_acc                     36\n","out_prncp                      7\n","out_prncp_inv                  7\n","total_pymnt                    7\n","total_pymnt_inv                7\n","total_rec_prncp                7\n","total_rec_int                  7\n","total_rec_late_fee             7\n","recoveries                     7\n","collection_recovery_fee        7\n","last_pymnt_amnt                7\n","dtype: int64"]},"metadata":{"tags":[]},"execution_count":10}]},{"cell_type":"code","metadata":{"id":"ryzrINVuze52","executionInfo":{"status":"ok","timestamp":1620243377013,"user_tz":240,"elapsed":9974,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Impute NaNs with mean \n","fillWithMean = ['loan_amnt','funded_amnt','funded_amnt_inv','term', \\\n","                'int_rate','installment','emp_length','annual_inc',\\\n","                'dti','open_acc','revol_bal','revol_util','total_acc',\\\n","                'out_prncp','out_prncp_inv','total_pymnt', \\\n","                'total_pymnt_inv','total_rec_prncp','total_rec_int', \\\n","                'last_pymnt_amnt']\n","\n","# Impute NaNs with zero\n","fillWithZero = ['delinq_2yrs','mths_since_last_delinq', \\\n","                'mths_since_last_record','pub_rec','total_rec_late_fee', \\\n","                'recoveries','collection_recovery_fee']\n","\n","# Perform imputation\n","im = impute.SimpleImputer(strategy='mean')   \n","data.loc[:,fillWithMean] = im.fit_transform(data[fillWithMean])\n","\n","data.loc[:,fillWithZero] = data.loc[:,fillWithZero].fillna(value=0,axis=1)"],"execution_count":11,"outputs":[]},{"cell_type":"code","metadata":{"id":"P0aHHMMAze52","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243377014,"user_tz":240,"elapsed":9972,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"dd3247ef-2346-4c17-9830-b5f0602ebf16"},"source":["# Check for NaNs one last time\n","nanCounter = np.isnan(data.loc[:,numericalFeats]).sum()\n","nanCounter"],"execution_count":12,"outputs":[{"output_type":"execute_result","data":{"text/plain":["loan_amnt                  0\n","funded_amnt                0\n","funded_amnt_inv            0\n","term                       0\n","int_rate                   0\n","installment                0\n","emp_length                 0\n","annual_inc                 0\n","dti                        0\n","delinq_2yrs                0\n","mths_since_last_delinq     0\n","mths_since_last_record     0\n","open_acc                   0\n","pub_rec                    0\n","revol_bal                  0\n","revol_util                 0\n","total_acc                  0\n","out_prncp                  0\n","out_prncp_inv              0\n","total_pymnt                0\n","total_pymnt_inv            0\n","total_rec_prncp            0\n","total_rec_int              0\n","total_rec_late_fee         0\n","recoveries                 0\n","collection_recovery_fee    0\n","last_pymnt_amnt            0\n","dtype: int64"]},"metadata":{"tags":[]},"execution_count":12}]},{"cell_type":"code","metadata":{"id":"h9Jqs5W0ze53","executionInfo":{"status":"ok","timestamp":1620243377014,"user_tz":240,"elapsed":9964,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Feature engineering\n","data['installmentOverLoanAmnt'] = data.installment/data.loan_amnt\n","data['loanAmntOverIncome'] = data.loan_amnt/data.annual_inc\n","data['revol_balOverIncome'] = data.revol_bal/data.annual_inc\n","data['totalPymntOverIncome'] = data.total_pymnt/data.annual_inc\n","data['totalPymntInvOverIncome'] = data.total_pymnt_inv/data.annual_inc\n","data['totalRecPrncpOverIncome'] = data.total_rec_prncp/data.annual_inc\n","data['totalRecIncOverIncome'] = data.total_rec_int/data.annual_inc\n","\n","newFeats = ['installmentOverLoanAmnt','loanAmntOverIncome', \\\n","            'revol_balOverIncome','totalPymntOverIncome', \\\n","           'totalPymntInvOverIncome','totalRecPrncpOverIncome', \\\n","            'totalRecIncOverIncome']"],"execution_count":13,"outputs":[]},{"cell_type":"code","metadata":{"id":"HM6AkTfpze53","executionInfo":{"status":"ok","timestamp":1620243377151,"user_tz":240,"elapsed":10099,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Select features for training\n","numericalPlusNewFeats = numericalFeats+newFeats\n","X_train = data.loc[:,numericalPlusNewFeats]\n","\n","# Scale data\n","sX = pp.StandardScaler()\n","X_train.loc[:,:] = sX.fit_transform(X_train)"],"execution_count":14,"outputs":[]},{"cell_type":"code","metadata":{"id":"8Kk9iqCDze53","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243377151,"user_tz":240,"elapsed":10097,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"0497ed7c-9f8a-4095-fd5a-cc6f66979756"},"source":["# View columns\n","X_train.columns"],"execution_count":15,"outputs":[{"output_type":"execute_result","data":{"text/plain":["Index(['loan_amnt', 'funded_amnt', 'funded_amnt_inv', 'term', 'int_rate',\n","       'installment', 'emp_length', 'annual_inc', 'dti', 'delinq_2yrs',\n","       'mths_since_last_delinq', 'mths_since_last_record', 'open_acc',\n","       'pub_rec', 'revol_bal', 'revol_util', 'total_acc', 'out_prncp',\n","       'out_prncp_inv', 'total_pymnt', 'total_pymnt_inv', 'total_rec_prncp',\n","       'total_rec_int', 'total_rec_late_fee', 'recoveries',\n","       'collection_recovery_fee', 'last_pymnt_amnt', 'installmentOverLoanAmnt',\n","       'loanAmntOverIncome', 'revol_balOverIncome', 'totalPymntOverIncome',\n","       'totalPymntInvOverIncome', 'totalRecPrncpOverIncome',\n","       'totalRecIncOverIncome'],\n","      dtype='object')"]},"metadata":{"tags":[]},"execution_count":15}]},{"cell_type":"code","metadata":{"id":"FQRBTlLUze53","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243377152,"user_tz":240,"elapsed":10091,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"a2783d50-26bc-4ee5-c5f8-62135fb34a33"},"source":["# Designate labels for evaluation\n","labels = data.grade\n","labels.unique()"],"execution_count":16,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array(['B', 'C', 'A', 'E', 'F', 'D', 'G', nan], dtype=object)"]},"metadata":{"tags":[]},"execution_count":16}]},{"cell_type":"code","metadata":{"id":"_dPkgMW2ze54","executionInfo":{"status":"ok","timestamp":1620243377152,"user_tz":240,"elapsed":10083,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Fill missing labels\n","labels = labels.fillna(value=\"Z\")\n","\n","# Convert labels to numerical values\n","lbl = pp.LabelEncoder()\n","lbl.fit(list(labels.values))\n","labels = pd.Series(data=lbl.transform(labels.values), name=\"grade\")\n","\n","# Store as y_train\n","y_train = labels"],"execution_count":17,"outputs":[]},{"cell_type":"code","metadata":{"id":"pXPAXk_Sze54","scrolled":true,"colab":{"base_uri":"https://localhost:8080/","height":427},"executionInfo":{"status":"ok","timestamp":1620243377237,"user_tz":240,"elapsed":10165,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"7e3e77bd-865c-4a3a-fc85-3180f17cccd2"},"source":["# Show numerical labels\n","labelsOriginalVSNew = pd.concat([labels, data.grade],axis=1)\n","labelsOriginalVSNew"],"execution_count":18,"outputs":[{"output_type":"execute_result","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>grade</th>\n","      <th>grade</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>1</td>\n","      <td>B</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>2</td>\n","      <td>C</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>2</td>\n","      <td>C</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>2</td>\n","      <td>C</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>1</td>\n","      <td>B</td>\n","    </tr>\n","    <tr>\n","      <th>...</th>\n","      <td>...</td>\n","      <td>...</td>\n","    </tr>\n","    <tr>\n","      <th>42537</th>\n","      <td>0</td>\n","      <td>A</td>\n","    </tr>\n","    <tr>\n","      <th>42538</th>\n","      <td>7</td>\n","      <td>NaN</td>\n","    </tr>\n","    <tr>\n","      <th>42539</th>\n","      <td>7</td>\n","      <td>NaN</td>\n","    </tr>\n","    <tr>\n","      <th>42540</th>\n","      <td>7</td>\n","      <td>NaN</td>\n","    </tr>\n","    <tr>\n","      <th>42541</th>\n","      <td>7</td>\n","      <td>NaN</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>42542 rows × 2 columns</p>\n","</div>"],"text/plain":["      grade grade\n","0         1     B\n","1         2     C\n","2         2     C\n","3         2     C\n","4         1     B\n","...     ...   ...\n","42537     0     A\n","42538     7   NaN\n","42539     7   NaN\n","42540     7   NaN\n","42541     7   NaN\n","\n","[42542 rows x 2 columns]"]},"metadata":{"tags":[]},"execution_count":18}]},{"cell_type":"code","metadata":{"id":"qGG74Ub5ze54","colab":{"base_uri":"https://localhost:8080/","height":334},"executionInfo":{"status":"ok","timestamp":1620243378639,"user_tz":240,"elapsed":11559,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"5887e741-8506-4803-c341-fbede3d67c72"},"source":["# Compare loan grades with interest rates\n","interestAndGrade = pd.DataFrame(data=[data.int_rate,labels])\n","interestAndGrade = interestAndGrade.T\n","\n","interestAndGrade.groupby(\"grade\").mean()"],"execution_count":19,"outputs":[{"output_type":"execute_result","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>int_rate</th>\n","    </tr>\n","    <tr>\n","      <th>grade</th>\n","      <th></th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0.0</th>\n","      <td>734.270844</td>\n","    </tr>\n","    <tr>\n","      <th>1.0</th>\n","      <td>1101.420857</td>\n","    </tr>\n","    <tr>\n","      <th>2.0</th>\n","      <td>1349.988902</td>\n","    </tr>\n","    <tr>\n","      <th>3.0</th>\n","      <td>1557.714927</td>\n","    </tr>\n","    <tr>\n","      <th>4.0</th>\n","      <td>1737.676783</td>\n","    </tr>\n","    <tr>\n","      <th>5.0</th>\n","      <td>1926.530361</td>\n","    </tr>\n","    <tr>\n","      <th>6.0</th>\n","      <td>2045.125000</td>\n","    </tr>\n","    <tr>\n","      <th>7.0</th>\n","      <td>1216.501563</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["          int_rate\n","grade             \n","0.0     734.270844\n","1.0    1101.420857\n","2.0    1349.988902\n","3.0    1557.714927\n","4.0    1737.676783\n","5.0    1926.530361\n","6.0    2045.125000\n","7.0    1216.501563"]},"metadata":{"tags":[]},"execution_count":19}]},{"cell_type":"markdown","metadata":{"id":"MeRb6W5P9pv1"},"source":["# Define Function"]},{"cell_type":"code","metadata":{"id":"gPJOcuM1ze55","executionInfo":{"status":"ok","timestamp":1620243378639,"user_tz":240,"elapsed":11551,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# To Analyze Goodness of the Clusters\n","def analyzeCluster(clusterDF, labelsDF):\n","    countByCluster = \\\n","        pd.DataFrame(data=clusterDF['cluster'].value_counts())\n","    countByCluster.reset_index(inplace=True,drop=False)\n","    countByCluster.columns = ['cluster','clusterCount']\n","        \n","    preds = pd.concat([labelsDF,clusterDF], axis=1)\n","    preds.columns = ['trueLabel','cluster']\n","    \n","    countByLabel = pd.DataFrame(data=preds.groupby('trueLabel').count())\n","        \n","    countMostFreq = pd.DataFrame(data=preds.groupby('cluster').agg( \\\n","        lambda x:x.value_counts().iloc[0]))\n","    countMostFreq.reset_index(inplace=True,drop=False)\n","    countMostFreq.columns = ['cluster','countMostFrequent']\n","    \n","    accuracyDF = countMostFreq.merge(countByCluster, \\\n","        left_on=\"cluster\",right_on=\"cluster\")\n","    \n","    overallAccuracy = accuracyDF.countMostFrequent.sum()/ \\\n","        accuracyDF.clusterCount.sum()\n","    \n","    accuracyByLabel = accuracyDF.countMostFrequent/ \\\n","        accuracyDF.clusterCount\n","    \n","    return countByCluster, countByLabel, countMostFreq, \\\n","        accuracyDF, overallAccuracy, accuracyByLabel"],"execution_count":20,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"x5Pmp8thbeWG"},"source":["# Dimensionality Reduction"]},{"cell_type":"code","metadata":{"id":"wDBFMzfRbgEc","executionInfo":{"status":"ok","timestamp":1620243378760,"user_tz":240,"elapsed":11671,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Principal Component Analysis\n","from sklearn.decomposition import PCA\n","\n","n_components = 34\n","whiten = False\n","random_state = 2018\n","\n","pca = PCA(n_components=n_components, whiten=whiten, \\\n","          random_state=random_state)\n","\n","X_train_PCA = pca.fit_transform(X_train)\n","X_train_PCA = pd.DataFrame(data=X_train_PCA, index=X_train.index)"],"execution_count":21,"outputs":[]},{"cell_type":"code","metadata":{"id":"WCSJrG-abrv3","executionInfo":{"status":"ok","timestamp":1620243378760,"user_tz":240,"elapsed":11669,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":[" # Log data\n","from sklearn.utils import resample\n","cwd = os.getcwd()\n","log_dir = cwd+\"/logs/06_group_segmentation/\"\n","sample = resample(y_train,replace=False,n_samples=2000,random_state=2021,stratify=y_train)\n","sample.to_csv(log_dir+'labels.tsv', sep = '\\t', index=False, header=False)"],"execution_count":22,"outputs":[]},{"cell_type":"code","metadata":{"id":"kpV2kgTRwpEE","executionInfo":{"status":"ok","timestamp":1620243378860,"user_tz":240,"elapsed":11768,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Write dimensions to CSV\n","X_train_PCA.loc[sample.index,0:3].to_csv(log_dir+'pca_data.tsv', sep = '\\t', index=False, header=False)"],"execution_count":23,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"bu2TupzC9w_C"},"source":["# K-means Application"]},{"cell_type":"code","metadata":{"id":"eSMO_qheze55","executionInfo":{"status":"ok","timestamp":1620243776453,"user_tz":240,"elapsed":409359,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Apply K-Means\n","from sklearn.cluster import KMeans\n","\n","n_clusters = 10\n","n_init = 10\n","max_iter = 300\n","tol = 0.0001\n","random_state = 2018\n","\n","kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n","                max_iter=max_iter, tol=tol, \\\n","                random_state=random_state)\n","\n","kMeans_inertia = pd.DataFrame(data=[],index=range(10,31), \\\n","                              columns=['inertia'])\n","\n","overallAccuracy_kMeansDF = pd.DataFrame(data=[], \\\n","    index=range(10,31),columns=['overallAccuracy'])\n","\n","for n_clusters in range(10,31):\n","    kmeans = KMeans(n_clusters=n_clusters, n_init=n_init, \\\n","                    max_iter=max_iter, tol=tol, \\\n","                    random_state=random_state)\n","\n","    kmeans.fit(X_train_PCA)\n","    kMeans_inertia.loc[n_clusters] = kmeans.inertia_\n","    X_train_kmeansClustered = kmeans.predict(X_train_PCA)\n","    X_train_kmeansClustered = pd.DataFrame(data= \\\n","        X_train_kmeansClustered, index=X_train.index, \\\n","        columns=['cluster'])\n","    \n","    countByCluster_kMeans, countByLabel_kMeans, \\\n","    countMostFreq_kMeans, accuracyDF_kMeans, \\\n","    overallAccuracy_kMeans, accuracyByLabel_kMeans = \\\n","    analyzeCluster(X_train_kmeansClustered, y_train)\n","    \n","    overallAccuracy_kMeansDF.loc[n_clusters] = \\\n","        overallAccuracy_kMeans"],"execution_count":24,"outputs":[]},{"cell_type":"code","metadata":{"id":"s0pVrIDCze56","colab":{"base_uri":"https://localhost:8080/","height":283},"executionInfo":{"status":"ok","timestamp":1620243776783,"user_tz":240,"elapsed":409687,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"c019b4bc-cc1c-495e-9f98-00fa4c522201"},"source":["# View plot\n","overallAccuracy_kMeansDF.plot()"],"execution_count":25,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.axes._subplots.AxesSubplot at 0x7fa7f3966bd0>"]},"metadata":{"tags":[]},"execution_count":25},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"ezwMaAyGze56","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243776783,"user_tz":240,"elapsed":409679,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"bfe0a011-1704-43e8-afe7-8a6804f6398f"},"source":["# View accuracy by cluster\n","accuracyByLabel_kMeans"],"execution_count":26,"outputs":[{"output_type":"execute_result","data":{"text/plain":["0     0.328299\n","1     0.261119\n","2     0.360078\n","3     0.232400\n","4     0.391180\n","5     0.326309\n","6     0.303797\n","7     0.543689\n","8     0.222222\n","9     0.401015\n","10    0.294190\n","11    0.312387\n","12    0.206897\n","13    0.315254\n","14    0.340435\n","15    0.711728\n","16    0.326621\n","17    0.365280\n","18    0.234783\n","19    0.289100\n","20    0.500000\n","21    0.374230\n","22    0.323799\n","23    0.250000\n","24    0.324701\n","25    0.231917\n","26    0.572813\n","27    0.259229\n","28    0.382329\n","29    0.269841\n","dtype: float64"]},"metadata":{"tags":[]},"execution_count":26}]},{"cell_type":"code","metadata":{"id":"cMsXioCktQt6","executionInfo":{"status":"ok","timestamp":1620243776783,"user_tz":240,"elapsed":409671,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Save cluster labels\n","X_train_kmeansClustered.loc[sample.index].to_csv(log_dir+'kmeans_cluster_labels.tsv', sep = '\\t', index=False, header=False)"],"execution_count":27,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"YSa_4F6b98_V"},"source":["# Hierarchical Clustering Application"]},{"cell_type":"code","metadata":{"id":"3xx3lOKgze56","executionInfo":{"status":"ok","timestamp":1620243879375,"user_tz":240,"elapsed":512261,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Hierarchical Clustering Application\n","import fastcluster\n","from scipy.cluster.hierarchy import dendrogram\n","from scipy.cluster.hierarchy import cophenet\n","from scipy.spatial.distance import pdist\n","\n","Z = fastcluster.linkage_vector(X_train_PCA, method='ward', \\\n","                               metric='euclidean')\n","\n","Z_dataFrame = pd.DataFrame(data=Z,columns=['clusterOne', \\\n","                'clusterTwo','distance','newClusterSize'])"],"execution_count":28,"outputs":[]},{"cell_type":"code","metadata":{"id":"xvM2XQz0ze57","colab":{"base_uri":"https://localhost:8080/","height":682},"executionInfo":{"status":"ok","timestamp":1620243879376,"user_tz":240,"elapsed":512260,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"c67db67b-8706-418d-a85a-9f56c4bfa4fb"},"source":["# View bottom most leaves\n","Z_dataFrame[:20]"],"execution_count":29,"outputs":[{"output_type":"execute_result","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>clusterOne</th>\n","      <th>clusterTwo</th>\n","      <th>distance</th>\n","      <th>newClusterSize</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>39786.0</td>\n","      <td>39787.0</td>\n","      <td>0.000000e+00</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>39788.0</td>\n","      <td>42542.0</td>\n","      <td>0.000000e+00</td>\n","      <td>3.0</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>42540.0</td>\n","      <td>42541.0</td>\n","      <td>0.000000e+00</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>42538.0</td>\n","      <td>42539.0</td>\n","      <td>5.602756e-17</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>42544.0</td>\n","      <td>42545.0</td>\n","      <td>1.318113e-16</td>\n","      <td>4.0</td>\n","    </tr>\n","    <tr>\n","      <th>5</th>\n","      <td>42543.0</td>\n","      <td>42546.0</td>\n","      <td>2.849554e-16</td>\n","      <td>7.0</td>\n","    </tr>\n","    <tr>\n","      <th>6</th>\n","      <td>33251.0</td>\n","      <td>33261.0</td>\n","      <td>1.561313e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>7</th>\n","      <td>42512.0</td>\n","      <td>42535.0</td>\n","      <td>3.342654e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>8</th>\n","      <td>42219.0</td>\n","      <td>42316.0</td>\n","      <td>3.368231e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>9</th>\n","      <td>6112.0</td>\n","      <td>21928.0</td>\n","      <td>3.384368e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>10</th>\n","      <td>33248.0</td>\n","      <td>33275.0</td>\n","      <td>3.583819e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>11</th>\n","      <td>33253.0</td>\n","      <td>33265.0</td>\n","      <td>3.595331e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>12</th>\n","      <td>33258.0</td>\n","      <td>42552.0</td>\n","      <td>3.719377e-01</td>\n","      <td>3.0</td>\n","    </tr>\n","    <tr>\n","      <th>13</th>\n","      <td>20430.0</td>\n","      <td>23299.0</td>\n","      <td>3.757307e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>14</th>\n","      <td>5455.0</td>\n","      <td>32845.0</td>\n","      <td>3.828709e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>15</th>\n","      <td>28615.0</td>\n","      <td>30306.0</td>\n","      <td>3.900294e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>16</th>\n","      <td>9056.0</td>\n","      <td>9769.0</td>\n","      <td>3.967378e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>17</th>\n","      <td>11162.0</td>\n","      <td>13857.0</td>\n","      <td>3.991124e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","    <tr>\n","      <th>18</th>\n","      <td>33270.0</td>\n","      <td>42548.0</td>\n","      <td>3.995620e-01</td>\n","      <td>3.0</td>\n","    </tr>\n","    <tr>\n","      <th>19</th>\n","      <td>17422.0</td>\n","      <td>17986.0</td>\n","      <td>4.061704e-01</td>\n","      <td>2.0</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["    clusterOne  clusterTwo      distance  newClusterSize\n","0      39786.0     39787.0  0.000000e+00             2.0\n","1      39788.0     42542.0  0.000000e+00             3.0\n","2      42540.0     42541.0  0.000000e+00             2.0\n","3      42538.0     42539.0  5.602756e-17             2.0\n","4      42544.0     42545.0  1.318113e-16             4.0\n","5      42543.0     42546.0  2.849554e-16             7.0\n","6      33251.0     33261.0  1.561313e-01             2.0\n","7      42512.0     42535.0  3.342654e-01             2.0\n","8      42219.0     42316.0  3.368231e-01             2.0\n","9       6112.0     21928.0  3.384368e-01             2.0\n","10     33248.0     33275.0  3.583819e-01             2.0\n","11     33253.0     33265.0  3.595331e-01             2.0\n","12     33258.0     42552.0  3.719377e-01             3.0\n","13     20430.0     23299.0  3.757307e-01             2.0\n","14      5455.0     32845.0  3.828709e-01             2.0\n","15     28615.0     30306.0  3.900294e-01             2.0\n","16      9056.0      9769.0  3.967378e-01             2.0\n","17     11162.0     13857.0  3.991124e-01             2.0\n","18     33270.0     42548.0  3.995620e-01             3.0\n","19     17422.0     17986.0  4.061704e-01             2.0"]},"metadata":{"tags":[]},"execution_count":29}]},{"cell_type":"code","metadata":{"id":"Hu6X7Fcpze57","colab":{"base_uri":"https://localhost:8080/","height":682},"executionInfo":{"status":"ok","timestamp":1620243879377,"user_tz":240,"elapsed":512253,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"9d9f72f6-434e-45e9-9c8b-110e8f888d8e"},"source":["# View top most leaves\n","Z_dataFrame[42521:]"],"execution_count":30,"outputs":[{"output_type":"execute_result","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>clusterOne</th>\n","      <th>clusterTwo</th>\n","      <th>distance</th>\n","      <th>newClusterSize</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>42521</th>\n","      <td>85038.0</td>\n","      <td>85043.0</td>\n","      <td>132.715723</td>\n","      <td>3969.0</td>\n","    </tr>\n","    <tr>\n","      <th>42522</th>\n","      <td>85051.0</td>\n","      <td>85052.0</td>\n","      <td>141.386569</td>\n","      <td>2899.0</td>\n","    </tr>\n","    <tr>\n","      <th>42523</th>\n","      <td>85026.0</td>\n","      <td>85027.0</td>\n","      <td>146.976703</td>\n","      <td>2351.0</td>\n","    </tr>\n","    <tr>\n","      <th>42524</th>\n","      <td>85048.0</td>\n","      <td>85049.0</td>\n","      <td>152.660192</td>\n","      <td>5691.0</td>\n","    </tr>\n","    <tr>\n","      <th>42525</th>\n","      <td>85036.0</td>\n","      <td>85059.0</td>\n","      <td>153.512281</td>\n","      <td>5956.0</td>\n","    </tr>\n","    <tr>\n","      <th>42526</th>\n","      <td>85033.0</td>\n","      <td>85044.0</td>\n","      <td>160.825959</td>\n","      <td>2203.0</td>\n","    </tr>\n","    <tr>\n","      <th>42527</th>\n","      <td>85055.0</td>\n","      <td>85061.0</td>\n","      <td>163.701428</td>\n","      <td>668.0</td>\n","    </tr>\n","    <tr>\n","      <th>42528</th>\n","      <td>85062.0</td>\n","      <td>85066.0</td>\n","      <td>168.199295</td>\n","      <td>6897.0</td>\n","    </tr>\n","    <tr>\n","      <th>42529</th>\n","      <td>85054.0</td>\n","      <td>85060.0</td>\n","      <td>168.924039</td>\n","      <td>9414.0</td>\n","    </tr>\n","    <tr>\n","      <th>42530</th>\n","      <td>85028.0</td>\n","      <td>85064.0</td>\n","      <td>185.215769</td>\n","      <td>3118.0</td>\n","    </tr>\n","    <tr>\n","      <th>42531</th>\n","      <td>85067.0</td>\n","      <td>85071.0</td>\n","      <td>187.832588</td>\n","      <td>15370.0</td>\n","    </tr>\n","    <tr>\n","      <th>42532</th>\n","      <td>85056.0</td>\n","      <td>85073.0</td>\n","      <td>203.212147</td>\n","      <td>17995.0</td>\n","    </tr>\n","    <tr>\n","      <th>42533</th>\n","      <td>85057.0</td>\n","      <td>85063.0</td>\n","      <td>205.285993</td>\n","      <td>9221.0</td>\n","    </tr>\n","    <tr>\n","      <th>42534</th>\n","      <td>85068.0</td>\n","      <td>85072.0</td>\n","      <td>207.902660</td>\n","      <td>5321.0</td>\n","    </tr>\n","    <tr>\n","      <th>42535</th>\n","      <td>85069.0</td>\n","      <td>85075.0</td>\n","      <td>236.754581</td>\n","      <td>9889.0</td>\n","    </tr>\n","    <tr>\n","      <th>42536</th>\n","      <td>85070.0</td>\n","      <td>85077.0</td>\n","      <td>298.587755</td>\n","      <td>16786.0</td>\n","    </tr>\n","    <tr>\n","      <th>42537</th>\n","      <td>85058.0</td>\n","      <td>85078.0</td>\n","      <td>309.946867</td>\n","      <td>16875.0</td>\n","    </tr>\n","    <tr>\n","      <th>42538</th>\n","      <td>85074.0</td>\n","      <td>85079.0</td>\n","      <td>375.698458</td>\n","      <td>34870.0</td>\n","    </tr>\n","    <tr>\n","      <th>42539</th>\n","      <td>85065.0</td>\n","      <td>85080.0</td>\n","      <td>400.711547</td>\n","      <td>37221.0</td>\n","    </tr>\n","    <tr>\n","      <th>42540</th>\n","      <td>85076.0</td>\n","      <td>85081.0</td>\n","      <td>644.047472</td>\n","      <td>42542.0</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["       clusterOne  clusterTwo    distance  newClusterSize\n","42521     85038.0     85043.0  132.715723          3969.0\n","42522     85051.0     85052.0  141.386569          2899.0\n","42523     85026.0     85027.0  146.976703          2351.0\n","42524     85048.0     85049.0  152.660192          5691.0\n","42525     85036.0     85059.0  153.512281          5956.0\n","42526     85033.0     85044.0  160.825959          2203.0\n","42527     85055.0     85061.0  163.701428           668.0\n","42528     85062.0     85066.0  168.199295          6897.0\n","42529     85054.0     85060.0  168.924039          9414.0\n","42530     85028.0     85064.0  185.215769          3118.0\n","42531     85067.0     85071.0  187.832588         15370.0\n","42532     85056.0     85073.0  203.212147         17995.0\n","42533     85057.0     85063.0  205.285993          9221.0\n","42534     85068.0     85072.0  207.902660          5321.0\n","42535     85069.0     85075.0  236.754581          9889.0\n","42536     85070.0     85077.0  298.587755         16786.0\n","42537     85058.0     85078.0  309.946867         16875.0\n","42538     85074.0     85079.0  375.698458         34870.0\n","42539     85065.0     85080.0  400.711547         37221.0\n","42540     85076.0     85081.0  644.047472         42542.0"]},"metadata":{"tags":[]},"execution_count":30}]},{"cell_type":"code","metadata":{"id":"k-4vYPndze57","executionInfo":{"status":"ok","timestamp":1620243879508,"user_tz":240,"elapsed":512376,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Cluster results\n","from scipy.cluster.hierarchy import fcluster\n","\n","distance_threshold = 100\n","clusters = fcluster(Z, distance_threshold, criterion='distance')\n","X_train_hierClustered = pd.DataFrame(data=clusters, \\\n","    index=X_train.index,columns=['cluster'])"],"execution_count":31,"outputs":[]},{"cell_type":"code","metadata":{"id":"XBM17FxIze58","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243879668,"user_tz":240,"elapsed":512534,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"4271a428-8061-4685-9b30-ba42e7c285a0"},"source":["# Show number of clusters created\n","print(\"Number of distinct clusters: \", \\\n","      len(X_train_hierClustered['cluster'].unique()))"],"execution_count":32,"outputs":[{"output_type":"stream","text":["Number of distinct clusters:  32\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"sp6HOwgNze58","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243879668,"user_tz":240,"elapsed":512527,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"105311b5-a2b4-439b-fa33-5c31a9886ac3"},"source":["# Show overall accuracy\n","countByCluster_hierClust, countByLabel_hierClust, \\\n","    countMostFreq_hierClust, accuracyDF_hierClust, \\\n","    overallAccuracy_hierClust, accuracyByLabel_hierClust = \\\n","    analyzeCluster(X_train_hierClustered, y_train)\n","\n","print(\"Overall accuracy from hierarchical clustering: \", \\\n","      overallAccuracy_hierClust)"],"execution_count":33,"outputs":[{"output_type":"stream","text":["Overall accuracy from hierarchical clustering:  0.3651685393258427\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"z5s_tRfRze58","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243879669,"user_tz":240,"elapsed":512520,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"78e6a993-fc64-423d-eeb5-a6dd95da6bc5"},"source":["# Show accuracy by cluster\n","print(\"Accuracy by cluster for hierarchical clustering\")\n","accuracyByLabel_hierClust"],"execution_count":34,"outputs":[{"output_type":"stream","text":["Accuracy by cluster for hierarchical clustering\n"],"name":"stdout"},{"output_type":"execute_result","data":{"text/plain":["0     0.304124\n","1     0.219001\n","2     0.228311\n","3     0.379722\n","4     0.240064\n","5     0.272011\n","6     0.314560\n","7     0.263930\n","8     0.246138\n","9     0.318942\n","10    0.302752\n","11    0.269772\n","12    0.335717\n","13    0.330403\n","14    0.346320\n","15    0.440141\n","16    0.744155\n","17    0.502227\n","18    0.294118\n","19    0.236111\n","20    0.254727\n","21    0.241042\n","22    0.317979\n","23    0.308771\n","24    0.284314\n","25    0.243243\n","26    0.500000\n","27    0.289157\n","28    0.365283\n","29    0.479693\n","30    0.393559\n","31    0.340875\n","dtype: float64"]},"metadata":{"tags":[]},"execution_count":34}]},{"cell_type":"code","metadata":{"id":"M2WvQCyMtkJ8","executionInfo":{"status":"ok","timestamp":1620243879669,"user_tz":240,"elapsed":512512,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Save cluster labels\n","X_train_hierClustered.loc[sample.index].to_csv(log_dir+'hierarchical_cluster_labels.tsv', sep = '\\t', index=False, header=False)"],"execution_count":35,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"cLWm5C3g-MfZ"},"source":["# HDBSCAN Application"]},{"cell_type":"code","metadata":{"id":"-Oj-5nTRze58","executionInfo":{"status":"ok","timestamp":1620243941443,"user_tz":240,"elapsed":574283,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# HDBSCAN Application\n","import hdbscan\n","\n","min_cluster_size = 20\n","min_samples = 20\n","alpha = 1.0\n","cluster_selection_method = 'leaf'\n","\n","hdb = hdbscan.HDBSCAN(min_cluster_size=min_cluster_size, \\\n","    min_samples=min_samples, alpha=alpha, \\\n","    cluster_selection_method=cluster_selection_method)\n","\n","X_train_hdbscanClustered = hdb.fit_predict(X_train_PCA)\n","X_train_hdbscanClustered = pd.DataFrame(data= \\\n","    X_train_hdbscanClustered, index=X_train.index, \\\n","    columns=['cluster'])\n","\n","countByCluster_hdbscan, countByLabel_hdbscan, \\\n","    countMostFreq_hdbscan, accuracyDF_hdbscan, \\\n","    overallAccuracy_hdbscan, accuracyByLabel_hdbscan = \\\n","    analyzeCluster(X_train_hdbscanClustered, y_train)"],"execution_count":36,"outputs":[]},{"cell_type":"code","metadata":{"id":"dfxd6ix0ze59","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243941446,"user_tz":240,"elapsed":574285,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"86ff8e9d-5ac5-4038-cd63-31e080e2129b"},"source":["# Show overall accuracy\n","print(\"Overall accuracy from HDBSCAN: \", overallAccuracy_hdbscan)"],"execution_count":37,"outputs":[{"output_type":"stream","text":["Overall accuracy from HDBSCAN:  0.3258426966292135\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"7w3PkuAMze59","colab":{"base_uri":"https://localhost:8080/","height":352},"executionInfo":{"status":"ok","timestamp":1620243941447,"user_tz":240,"elapsed":574284,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"09526fc7-85fb-4f31-e21d-9eeea7332285"},"source":["# Count by cluster\n","print(\"Cluster results for HDBSCAN\")\n","countByCluster_hdbscan"],"execution_count":38,"outputs":[{"output_type":"stream","text":["Cluster results for HDBSCAN\n"],"name":"stdout"},{"output_type":"execute_result","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>clusterCount</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>-1</td>\n","      <td>32478</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>6</td>\n","      <td>4261</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>2</td>\n","      <td>3668</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>1</td>\n","      <td>1096</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>4</td>\n","      <td>786</td>\n","    </tr>\n","    <tr>\n","      <th>5</th>\n","      <td>0</td>\n","      <td>120</td>\n","    </tr>\n","    <tr>\n","      <th>6</th>\n","      <td>7</td>\n","      <td>72</td>\n","    </tr>\n","    <tr>\n","      <th>7</th>\n","      <td>3</td>\n","      <td>38</td>\n","    </tr>\n","    <tr>\n","      <th>8</th>\n","      <td>5</td>\n","      <td>23</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["   cluster  clusterCount\n","0       -1         32478\n","1        6          4261\n","2        2          3668\n","3        1          1096\n","4        4           786\n","5        0           120\n","6        7            72\n","7        3            38\n","8        5            23"]},"metadata":{"tags":[]},"execution_count":38}]},{"cell_type":"code","metadata":{"id":"gknDIjlMze59","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1620243941447,"user_tz":240,"elapsed":574283,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"1122dd4e-7dc1-43cc-82ef-d9bbefaf56ee"},"source":["# Accuracy by cluster\n","accuracyByLabel_hdbscan"],"execution_count":39,"outputs":[{"output_type":"execute_result","data":{"text/plain":["0    0.284346\n","1    0.341667\n","2    0.414234\n","3    0.332061\n","4    0.552632\n","5    0.438931\n","6    0.391304\n","7    0.588829\n","8    0.416667\n","dtype: float64"]},"metadata":{"tags":[]},"execution_count":39}]},{"cell_type":"code","metadata":{"id":"09TK0QietwSe","executionInfo":{"status":"ok","timestamp":1620243941543,"user_tz":240,"elapsed":574377,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}}},"source":["# Save cluster labels\n","X_train_hdbscanClustered.loc[sample.index].to_csv(log_dir+'hdbscan_cluster_labels.tsv', sep = '\\t', index=False, header=False)"],"execution_count":40,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"xhBvF3JD-UDD"},"source":["# Conclusion"]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"cPLsneGKp4JG","executionInfo":{"status":"ok","timestamp":1620243941544,"user_tz":240,"elapsed":574377,"user":{"displayName":"Ankur Patel","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GjuyT9p2uQcITYiB6Ja4gdHeMPvvVZ_RNUvS_DFuY0=s64","userId":"08322043083056308084"}},"outputId":"a9af1aeb-094c-47cc-ebeb-9f1f8e0b28fa"},"source":["import datetime as dt\n","print(\"Completed: \", dt.datetime.now())"],"execution_count":41,"outputs":[{"output_type":"stream","text":["Completed:  2021-05-05 19:45:41.336670\n"],"name":"stdout"}]}]}