{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"import matplotlib.pyplot as plt\nfrom copy import deepcopy\nfrom tqdm import tqdm\nimport pandas as pd\nimport numpy as np\nimport warnings\nimport sys\nimport gc\n\nfrom sklearn import model_selection\nfrom catboost import CatBoostClassifier\nwarnings.filterwarnings('ignore')","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-13T07:22:25.847860Z","iopub.execute_input":"2022-06-13T07:22:25.848578Z","iopub.status.idle":"2022-06-13T07:22:27.407519Z","shell.execute_reply.started":"2022-06-13T07:22:25.848484Z","shell.execute_reply":"2022-06-13T07:22:27.406319Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
Introduction
","metadata":{}},{"cell_type":"markdown","source":"#####

If we look at the dataset, we can see there is one column in the datset corrosponding to statement date. In this notebook we will explore how other features has been changed from customers statement date. As an example we will explore how customer's balance features has changed over his/her deferent statement dates. Below you can see how statements distributed accross the train and test datasets. Image borrowed from -[ambrosm](https://www.kaggle.com/code/ambrosm/amex-eda-which-makes-sense)\n\n![image.png](attachment:125f6cf7-677d-43c7-9477-a9b433c967fe.png)\n\n**Methodlogy used in below figures.**\n\n#####

First read first 200K samples in training dataset. Then ranked each customers statement and declare the column representing statement sequance. Then analyzed how each feature changed over diferent statement dates. For both Non default and Default segmants I've used same instance count(200) when plotting","metadata":{},"attachments":{"125f6cf7-677d-43c7-9477-a9b433c967fe.png":{"image/png":"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"}}},{"cell_type":"code","source":"DEFAULT_COLOR = '#FF5F00'\nNON_DEFAULT_COLOR = '#003D93'\nTXT_BACK_ND = '#40C895'\nTXT_BACK_D = '#E37D1D'\nN = 200","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:23:51.358832Z","iopub.execute_input":"2022-06-13T07:23:51.359238Z","iopub.status.idle":"2022-06-13T07:23:51.364457Z","shell.execute_reply.started":"2022-06-13T07:23:51.359204Z","shell.execute_reply":"2022-06-13T07:23:51.363575Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"reader = pd.read_csv(\"../input/amex-default-prediction/train_data.csv\",chunksize=200_000,iterator=True)\nlabels = pd.read_csv('../input/amex-default-prediction/train_labels.csv',index_col='customer_ID')","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:23:51.752987Z","iopub.execute_input":"2022-06-13T07:23:51.753386Z","iopub.status.idle":"2022-06-13T07:23:52.813541Z","shell.execute_reply.started":"2022-06-13T07:23:51.753356Z","shell.execute_reply":"2022-06-13T07:23:52.812635Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df = next(reader)\nna_pct = df.isna().sum()/len(df)\ncols_to_drop = na_pct[na_pct>0.90].index.tolist()\n\ndf['S_2'] = pd.DatetimeIndex(df['S_2'])\ndf.drop(cols_to_drop,axis=1,inplace=True)\n\nall_feats = df.select_dtypes(exclude='object').columns.tolist()\n\ndf['date_rank'] = df.groupby(\"customer_ID\")[\"S_2\"].rank(\"dense\")","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:23:52.814973Z","iopub.execute_input":"2022-06-13T07:23:52.815328Z","iopub.status.idle":"2022-06-13T07:24:06.630064Z","shell.execute_reply.started":"2022-06-13T07:23:52.815299Z","shell.execute_reply":"2022-06-13T07:24:06.629214Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def plot_features_vs_time(features,name,figsize=(20,70)):\n N_COLS = 4\n N_ROWS = int(np.ceil(len(features)/2))\n\n fig,ax = plt.subplots(N_ROWS,N_COLS,figsize=figsize,constrained_layout=True)\n axi = ax.reshape(-1,2)\n\n for i,feature in enumerate(features):\n\n feat_df = df[[feature]+['customer_ID','date_rank']]\n feat_df = feat_df.pivot_table(index='customer_ID',columns='date_rank')[feature]\n feat_df = feat_df.join(labels)\n feat_df_1 = feat_df[feat_df.target==1]\n feat_df_0 = feat_df[feat_df.target==0]\n # print(i,feature)\n for x in range(N):\n axi[i][0].plot(feat_df_1.iloc[x,range(13)],color=DEFAULT_COLOR,alpha=x/N);\n axi[i][0].grid('True');\n axi[i][0].set_title(label=f'Feature:{feature} [Default]',fontweight='semibold',fontfamily='serif',backgroundcolor=TXT_BACK_D)\n\n axi[i][1].plot(feat_df_0.iloc[x,range(13)],color=NON_DEFAULT_COLOR,alpha=x/N);\n axi[i][1].grid('True');\n axi[i][1].set_title(label=f'Feature:{feature} [Non Default]',fontweight='semibold',fontfamily='serif',backgroundcolor=TXT_BACK_ND)\n\n fig.suptitle(f'{name} Over Time Period\\n\\n',fontweight='bold',fontfamily='serif',fontsize=19)\n # plt.tight_layout()\n plt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:24:06.631593Z","iopub.execute_input":"2022-06-13T07:24:06.631936Z","iopub.status.idle":"2022-06-13T07:24:06.644430Z","shell.execute_reply.started":"2022-06-13T07:24:06.631906Z","shell.execute_reply":"2022-06-13T07:24:06.643738Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#

1. Balance features vs. Statement date
","metadata":{}},{"cell_type":"code","source":"balance_features = [col for col in all_feats if col.startswith('B')]\nplot_features_vs_time(balance_features,name='Balance Features')","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:24:06.645318Z","iopub.execute_input":"2022-06-13T07:24:06.645740Z","iopub.status.idle":"2022-06-13T07:25:02.324878Z","shell.execute_reply.started":"2022-06-13T07:24:06.645713Z","shell.execute_reply":"2022-06-13T07:25:02.323594Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
1.1 Insights
","metadata":{}},{"cell_type":"markdown","source":"#####

As per the plots above, we can notice some significant insights.

\n* **B_4** - In **Default** users, ~90% of B_4 feature distributed between 0-1. This range differ for **Non Default** users which is 0-0.5 .\n* **B_6** - For **Default** users this feature range from 0.0 to ~0.25. For **Non Default** users it's from 0.00 to ~1.0.\n* **B_7** - For **Default** users this feature almost evenly distributed betwwen 0.0 and 1.0 For **Non Default** users this feature packed betwwen 0 and 0.1 .\n* **B_23** - Almost same behaviour as B_7. But range may differ.\n* **B_36** - As per plots, it seems there is significance deference in two cases. Have to re-check confirm it further since plots' scales are different.(Todo)\n* **B_41** - Seems there is signifinact diference. For **default** users upper and lower bundry is 1.00 and 0.00, for **Non Default** users this there is no clear upper boundry. The lower boundry is 0.00.(For Non Default users this feature almost 0)","metadata":{}},{"cell_type":"markdown","source":"#
2. Delinquency features vs. Statement date
","metadata":{}},{"cell_type":"code","source":"delinquent_features = [col for col in all_feats if col.startswith('D')]\nplot_features_vs_time(delinquent_features,name='Delinquent Features',figsize=(20,100))","metadata":{"execution":{"iopub.status.busy":"2022-06-13T09:41:25.137435Z","iopub.execute_input":"2022-06-13T09:41:25.137906Z","iopub.status.idle":"2022-06-13T09:41:45.384820Z","shell.execute_reply.started":"2022-06-13T09:41:25.137870Z","shell.execute_reply":"2022-06-13T09:41:45.383578Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
2.1 Insights
","metadata":{}},{"cell_type":"markdown","source":"#####

Being delinquent refers to the state of being past due on a debt. Delinquency occurs as soon as a borrower misses a payment on a loan, which can affect their credit score.
Ref- www.investopedia.com

\n\n* **D_48** - In **Default** users this value skewed towards to 0.5 to 1. In **Non Default** users majority values between 0.5 and 1.0 .\n* **D_54** - For **Default** users this feature has clear lower and upper boundries, respectively 0 & 1. For **Non Default** users this value is almost 1.0 .\n* **D_79** - For **Default** users this feature is 0,0.5,1 or 1.5 (Most cases), but for **Non Default** users this is differ.0.0, 0.5 and 1.0\n* **D_86 & D_127** - Kindof same **D_54** behaviour.","metadata":{}},{"cell_type":"markdown","source":"#
3. Risk features vs. Statement date
","metadata":{}},{"cell_type":"code","source":"risk_features = [col for col in all_feats if col.startswith('R')]\nplot_features_vs_time(risk_features,name='Risk Features',figsize=(20,30))","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:31:06.372066Z","iopub.execute_input":"2022-06-13T07:31:06.372867Z","iopub.status.idle":"2022-06-13T07:31:48.955982Z","shell.execute_reply.started":"2022-06-13T07:31:06.372830Z","shell.execute_reply":"2022-06-13T07:31:48.955002Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
3.1 Insights
","metadata":{}},{"cell_type":"markdown","source":"#####

In most cases we couldn't see clear signals that we can segregrate two bases. Found below tiny decision points.

","metadata":{}},{"cell_type":"markdown","source":" - **R_2 and R_4** - For **Non Default** users this value keep same (0) accross the statement dates.\n - **R_19 and R_21** - For **Non Default** users this value is almost 0","metadata":{}},{"cell_type":"markdown","source":"#
4. Payment features vs. Statement date
","metadata":{}},{"cell_type":"code","source":"payment_features = [col for col in all_feats if col.startswith('P')]\nplot_features_vs_time(payment_features,name='Payment Features',figsize=(20,8))","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:31:48.957424Z","iopub.execute_input":"2022-06-13T07:31:48.957737Z","iopub.status.idle":"2022-06-13T07:31:53.583464Z","shell.execute_reply.started":"2022-06-13T07:31:48.957704Z","shell.execute_reply":"2022-06-13T07:31:53.582602Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
4.1 Insights
","metadata":{}},{"cell_type":"markdown","source":"#####

For Payment feature, I could not notice clear signals.

","metadata":{}},{"cell_type":"markdown","source":"#
5. Spend features vs. Statement date
","metadata":{}},{"cell_type":"code","source":"spend_features = [col for col in all_feats if col.startswith('S')]\nspend_features.remove('S_2')\nplot_features_vs_time(spend_features,name='Spend Features',figsize=(20,30))","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:31:53.585801Z","iopub.execute_input":"2022-06-13T07:31:53.586368Z","iopub.status.idle":"2022-06-13T07:32:25.319113Z","shell.execute_reply.started":"2022-06-13T07:31:53.586329Z","shell.execute_reply":"2022-06-13T07:32:25.317291Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
5.1 Insights
","metadata":{}},{"cell_type":"markdown","source":"#####

Below are the noticeable signal in Payment features

\n\n* **S_22 and S_24** - For **Default** users this feature almost 0, and **Non Default** users this is more skewed towards to 1.0.","metadata":{}},{"cell_type":"code","source":"%whos DataFrame","metadata":{"execution":{"iopub.status.busy":"2022-06-13T07:32:25.327364Z","iopub.execute_input":"2022-06-13T07:32:25.327717Z","iopub.status.idle":"2022-06-13T07:32:25.395844Z","shell.execute_reply.started":"2022-06-13T07:32:25.327678Z","shell.execute_reply":"2022-06-13T07:32:25.394912Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"del df;del labels;gc.collect()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-13T07:32:25.397004Z","iopub.execute_input":"2022-06-13T07:32:25.397347Z","iopub.status.idle":"2022-06-13T07:32:26.385125Z","shell.execute_reply.started":"2022-06-13T07:32:25.397317Z","shell.execute_reply":"2022-06-13T07:32:26.384205Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#
6. Conclusion
","metadata":{}},{"cell_type":"markdown","source":"#####

Aim of this notebook it provide one way to understand given dataset. Above insight may differ for different customer sample. You may do your own research before using these insights for your modeling/Analysis. If you noticed any mistake above, please let me know. Glad to fix them.

","metadata":{}},{"cell_type":"markdown","source":"#
Thank you!
","metadata":{}}]}