{"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":"markdown","source":"![AMEX EDA banner.PNG](attachment:883dc28e-0c6d-46f2-af1a-c3117fe42958.PNG)","metadata":{},"attachments":{"883dc28e-0c6d-46f2-af1a-c3117fe42958.PNG":{"image/png":"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"}}},{"cell_type":"markdown","source":"# <b>1. Introduction</b>\n\nAmerican Express Company (called also Amex) is a multinational corporation with a sepcialisation in payment cards. It is present on the New York Stock Exchange (NYSE) with a ticker AXP. New York is also a city where they have the headquarter. They employ over 63k employees worldwide and hold about 23% payment card market in US.","metadata":{}},{"cell_type":"markdown","source":"![1.jpg](attachment:2c740be3-dcdd-4b26-b072-50d8054530af.jpg)","metadata":{},"attachments":{"2c740be3-dcdd-4b26-b072-50d8054530af.jpg":{"image/jpeg":"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"}}},{"cell_type":"markdown","source":"As for many banks the core of its operation is crediting, it's crucial to manage it efficiently. Especially banks want to lend money only to people who can pay the loan. That's why in this chalange we are asked to find out which customers are not able to do it. Not paying the money back is called `default`. This notebook explores the American Express (AMEX) churn prediction dataset.  \n\nHere's a nice [McKinsely's article](https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/grow-fast-or-die-slow-focusing-on-customer-success-to-drive-growth) about the impact of churn on businesses.\n\nSome other useful resources:\n* [Predict Customer Churn in Python (Towards Data Science)](https://towardsdatascience.com/predict-customer-churn-in-python-e8cd6d3aaa7)\n* [Churn Prediction- Commercial use of Data Science (Analytics Vidhya](https://www.analyticsvidhya.com/blog/2021/08/churn-prediction-commercial-use-of-data-science/)\n* [Can You Predict Customer Churn ? (Youtube)](https://www.youtube.com/watch?v=ocMd2loRfWE)\n\nIn this competition the evaluation metric is referenced in [this notebook](https://www.kaggle.com/code/inversion/amex-competition-metric-python). It uses two submetrics where one is a normalized Gini coefficient, a good and clean explanation of this metrics can be found [here](https://theblog.github.io/post/gini-coefficient-intuitive-explanation/).","metadata":{}},{"cell_type":"markdown","source":"# <b>2. Reading libraries</b>","metadata":{}},{"cell_type":"code","source":"import os\nimport warnings\n\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns\nimport matplotlib as mpl\nimport math\nimport gc\nimport pprint\n\npd.set_option('display.max_columns', None)\nwarnings.filterwarnings('ignore')","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2022-07-21T12:20:20.984608Z","iopub.execute_input":"2022-07-21T12:20:20.985437Z","iopub.status.idle":"2022-07-21T12:20:22.302961Z","shell.execute_reply.started":"2022-07-21T12:20:20.985340Z","shell.execute_reply":"2022-07-21T12:20:22.301925Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3. Reading data </b>\n\nBecause this dataset is huge and reading it directly from the .csv file consumes almost the entire memory we have to find a workaround. There are essentialy two options:\n* Read data chunk by chunk\n* Used a compressed dataset","metadata":{}},{"cell_type":"code","source":"TRAIN_DATA_PATH = \"../input/amex-default-prediction/train_data.csv\"","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:22.304669Z","iopub.execute_input":"2022-07-21T12:20:22.305213Z","iopub.status.idle":"2022-07-21T12:20:22.311653Z","shell.execute_reply.started":"2022-07-21T12:20:22.305168Z","shell.execute_reply":"2022-07-21T12:20:22.309924Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.1 Reading in chunks </b>","metadata":{}},{"cell_type":"code","source":"chunksize = 13000\ndf_train_raw_chunks = pd.read_csv(TRAIN_DATA_PATH, chunksize=chunksize)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:22.313212Z","iopub.execute_input":"2022-07-21T12:20:22.313687Z","iopub.status.idle":"2022-07-21T12:20:22.335883Z","shell.execute_reply.started":"2022-07-21T12:20:22.313643Z","shell.execute_reply":"2022-07-21T12:20:22.334817Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Everytime you run the cell below you'll get a new chunk.","metadata":{}},{"cell_type":"code","source":"df_train_raw_ch = df_train_raw_chunks.__next__()\ndf_train_raw_ch.head()","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:22.337817Z","iopub.execute_input":"2022-07-21T12:20:22.338871Z","iopub.status.idle":"2022-07-21T12:20:23.540945Z","shell.execute_reply.started":"2022-07-21T12:20:22.338824Z","shell.execute_reply":"2022-07-21T12:20:23.539624Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Now, we can do with this chunk all what we want. For example subsample a chunk only to data of a single customer.","metadata":{}},{"cell_type":"code","source":"sample_customer_id = np.random.choice(df_train_raw_ch['customer_ID'])\ncustomer_data_ex = df_train_raw_ch[df_train_raw_ch[\"customer_ID\"] == sample_customer_id]\ncustomer_data_ex.head()","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:23.542505Z","iopub.execute_input":"2022-07-21T12:20:23.542936Z","iopub.status.idle":"2022-07-21T12:20:23.776590Z","shell.execute_reply.started":"2022-07-21T12:20:23.542900Z","shell.execute_reply":"2022-07-21T12:20:23.775315Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"As this method is somehow usefull for EDA it desn't allow you to explore the entire distribution of data and thus the second option is preffered. So let's clean the memory.","metadata":{}},{"cell_type":"code","source":"del df_train_raw_chunks, df_train_raw_ch, customer_data_ex, sample_customer_id\ngc.collect()","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:23.777856Z","iopub.execute_input":"2022-07-21T12:20:23.778298Z","iopub.status.idle":"2022-07-21T12:20:23.924344Z","shell.execute_reply.started":"2022-07-21T12:20:23.778255Z","shell.execute_reply":"2022-07-21T12:20:23.921892Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.2 Reading a compressed dataset </b>\n\nThere is an pre-processed, lightweight version of this dataset (train data already merged with labels): [AMEX-Feather-Dataset](https://www.kaggle.com/datasets/munumbutt/amexfeather) (thank you [Munum](https://www.kaggle.com/munumbutt)!). This dataset is available in Apache's feather format (more about it you can read [here](https://arrow.apache.org/docs/python/feather.html)). There is also a parquet version ([Amex Competition Data in Parquet Format](https://www.kaggle.com/datasets/odins0n/amex-parquet)).  \n\nHowever, the one I suggest to use [is this one](https://www.kaggle.com/datasets/raddar/amex-data-integer-dtypes-parquet-format), which is already precleaned by [raddar](https://www.kaggle.com/raddar) from the artificial noise.\n\nBefore you can use it, you have to add it to your environment by clicking **\"+ Add data\"** button in the upper right corner of Kaggle's interface. ","metadata":{}},{"cell_type":"code","source":"train_raw = pd.read_parquet('../input/amex-data-integer-dtypes-parquet-format/train.parquet')","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:23.926477Z","iopub.execute_input":"2022-07-21T12:20:23.927021Z","iopub.status.idle":"2022-07-21T12:20:43.846342Z","shell.execute_reply.started":"2022-07-21T12:20:23.926975Z","shell.execute_reply":"2022-07-21T12:20:43.845371Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"labels = pd.read_csv('../input/amex-default-prediction/train_labels.csv')","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:43.850890Z","iopub.execute_input":"2022-07-21T12:20:43.851876Z","iopub.status.idle":"2022-07-21T12:20:45.602371Z","shell.execute_reply.started":"2022-07-21T12:20:43.851831Z","shell.execute_reply":"2022-07-21T12:20:45.601154Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"labels.head()","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:45.603916Z","iopub.execute_input":"2022-07-21T12:20:45.604342Z","iopub.status.idle":"2022-07-21T12:20:45.623222Z","shell.execute_reply.started":"2022-07-21T12:20:45.604292Z","shell.execute_reply":"2022-07-21T12:20:45.621302Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_raw = train_raw.merge(labels, left_on='customer_ID', right_on='customer_ID')","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:20:45.627799Z","iopub.execute_input":"2022-07-21T12:20:45.629099Z","iopub.status.idle":"2022-07-21T12:22:25.121990Z","shell.execute_reply.started":"2022-07-21T12:20:45.629054Z","shell.execute_reply":"2022-07-21T12:22:25.120392Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_raw.shape","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:22:25.124286Z","iopub.execute_input":"2022-07-21T12:22:25.124861Z","iopub.status.idle":"2022-07-21T12:22:25.134610Z","shell.execute_reply.started":"2022-07-21T12:22:25.124805Z","shell.execute_reply":"2022-07-21T12:22:25.133652Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 4. EDA </b>\n\nColumns in the dataset are divided by the organisers in the following groups:  \n**`D_*`:** Delinquency variables  \n**`S_*`:** Spend variables  \n**`P_*`:** Payment variables  \n**`B_*`:** Balance variables  \n**`R_*`:** Risk variables  \nFollowing features are categorical: `B_30`, `B_38`, `D_63`, `D_64`, `D_66`, `D_68`, `D_114`, `D_116`, `D_117`, `D_120`, `D_126`.\n\n**`S_2`:** contains a timestamp","metadata":{}},{"cell_type":"code","source":"train_raw['S_2'] = pd.to_datetime(train_raw['S_2'])","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:22:25.135975Z","iopub.execute_input":"2022-07-21T12:22:25.136334Z","iopub.status.idle":"2022-07-21T12:22:26.287877Z","shell.execute_reply.started":"2022-07-21T12:22:25.136304Z","shell.execute_reply":"2022-07-21T12:22:26.286739Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"categorical_features = ['B_30', 'B_38', 'D_63', 'D_64', 'D_66', 'D_68', 'D_114', 'D_116', 'D_117', 'D_120', 'D_126']\ntrain_raw[categorical_features] = train_raw[categorical_features].astype(\"category\")\ntrain_raw[categorical_features].dtypes","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:22:26.289281Z","iopub.execute_input":"2022-07-21T12:22:26.289906Z","iopub.status.idle":"2022-07-21T12:22:32.334996Z","shell.execute_reply.started":"2022-07-21T12:22:26.289862Z","shell.execute_reply":"2022-07-21T12:22:32.333825Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"mpl.rcParams.update(mpl.rcParamsDefault)\n\ndata = train_raw\ndf_dtypes = train_raw.dtypes.value_counts()\n\nfig = plt.figure(figsize=(5,2),facecolor='white')\n\nax = fig.add_subplot(1,1,1)\nfont = 'monospace'\nax.text(1, 0.8, \"Key figures\",color='black',fontsize=28, fontweight='bold', fontfamily=font, ha='center')\n\nax.text(0, 0.4, \"{:,d}\".format(data.shape[0]), color='#fcba03', fontsize=24, fontweight='bold', fontfamily=font, ha='center')\nax.text(0, 0.001, \"# of rows \\nin the dataset\",color='dimgrey',fontsize=15, fontweight='light', fontfamily=font,ha='center')\n\nax.text(0.6, 0.4, \"{}\".format(data.shape[1]), color='#fcba03', fontsize=24, fontweight='bold', fontfamily=font, ha='center')\nax.text(0.6, 0.001, \"# of features \\nin the dataset\",color='dimgrey',fontsize=15, fontweight='light', fontfamily=font,ha='center')\n\nax.text(1.2, 0.4, \"{}\".format(len(data.select_dtypes(np.number).columns)), color='#fcba03', fontsize=24, fontweight='bold', fontfamily=font, ha='center')\nax.text(1.2, 0.001, \"# of numeric columns \\nin the dataset\",color='dimgrey',fontsize=15, fontweight='light', fontfamily=font, ha='center')\n\nax.text(1.9, 0.4,\"{}\".format(len(data.select_dtypes('datetime').columns)), color='#fcba03', fontsize=24, fontweight='bold', fontfamily=font, ha='center')\nax.text(1.9, 0.001,\"# of datetime columns \\nin the dataset\",color='dimgrey',fontsize=15, fontweight='light', fontfamily=font,ha='center')\n\nax.set_yticklabels('')\nax.tick_params(axis='y',length=0)\nax.tick_params(axis='x',length=0)\nax.set_xticklabels('')\n\nfor direction in ['top','right','left','bottom']:\n    ax.spines[direction].set_visible(False)\n\nfig.subplots_adjust(top=0.9, bottom=0.2, left=0, hspace=1)\n\nfig.patch.set_linewidth(5)\nfig.patch.set_edgecolor('#346eeb')\nfig.patch.set_facecolor('#f6f6f6')\nax.set_facecolor('#f6f6f6')\n    \nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-07-21T12:22:32.336459Z","iopub.execute_input":"2022-07-21T12:22:32.336855Z","iopub.status.idle":"2022-07-21T12:22:33.996619Z","shell.execute_reply.started":"2022-07-21T12:22:32.336822Z","shell.execute_reply":"2022-07-21T12:22:33.995813Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Let's check now what are the date ranges in our database.","metadata":{}},{"cell_type":"code","source":"print(f'Train dates range is from {train_raw[\"S_2\"].min()} to {train_raw[\"S_2\"].max()}.')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-07-21T12:22:33.998393Z","iopub.execute_input":"2022-07-21T12:22:33.998854Z","iopub.status.idle":"2022-07-21T12:22:34.042887Z","shell.execute_reply.started":"2022-07-21T12:22:33.998812Z","shell.execute_reply":"2022-07-21T12:22:34.041892Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"We have slightly over one year of data.\n\n# <b> 3.1 Missing data </b>\n\nLet's see now how many missing values do we have.","metadata":{}},{"cell_type":"code","source":"tmp = train_raw.isna().sum().div(len(train_raw)).mul(100).sort_values(ascending=False)","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-07-21T12:22:34.044404Z","iopub.execute_input":"2022-07-21T12:22:34.044764Z","iopub.status.idle":"2022-07-21T12:22:36.322506Z","shell.execute_reply.started":"2022-07-21T12:22:34.044732Z","shell.execute_reply":"2022-07-21T12:22:36.321745Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.style.use('Solarize_Light2')\nfig, ax = plt.subplots(2,1, figsize=(25,10))\nsns.barplot(x=tmp[:100].index, y=tmp[:100].values, ax=ax[0])\nsns.barplot(x=tmp[100:].index, y=tmp[100:].values, ax=ax[1])\nax[0].set_ylabel(\"Percentage [%]\"), ax[1].set_ylabel(\"Percentage [%]\")\nax[0].tick_params(axis='x', rotation=90); ax[1].tick_params(axis='x', rotation=90)\nplt.suptitle(\"Amount of missing data\")\nplt.tight_layout()\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-07-21T12:22:36.324806Z","iopub.execute_input":"2022-07-21T12:22:36.325287Z","iopub.status.idle":"2022-07-21T12:22:39.298800Z","shell.execute_reply.started":"2022-07-21T12:22:36.325241Z","shell.execute_reply":"2022-07-21T12:22:39.297691Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"There is a significant number of features with a lot of missing data. Top10 columns with missing data are:","metadata":{}},{"cell_type":"code","source":"tmp.head(10)\n#print(list(tmp.index[:10]))","metadata":{"execution":{"iopub.status.busy":"2022-07-21T12:24:34.104629Z","iopub.execute_input":"2022-07-21T12:24:34.105194Z","iopub.status.idle":"2022-07-21T12:24:34.119358Z","shell.execute_reply.started":"2022-07-21T12:24:34.105147Z","shell.execute_reply":"2022-07-21T12:24:34.118625Z"},"_kg_hide-input":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.2 Distribution of a target variable </b>\n\nLet's check now the distribution of the `target` variable.","metadata":{}},{"cell_type":"code","source":"tmp = train_raw['target'].value_counts().div(len(train_raw)).mul(100)","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:17:42.733307Z","iopub.execute_input":"2022-06-09T17:17:42.733831Z","iopub.status.idle":"2022-06-09T17:17:42.768808Z","shell.execute_reply.started":"2022-06-09T17:17:42.733799Z","shell.execute_reply":"2022-06-09T17:17:42.767718Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ax = sns.barplot(x=tmp.index, y=tmp.values)\nax.bar_label(ax.containers[0], fmt='%.f%%')\nplt.title(\"Distribution of a target variable\")\nplt.ylabel(\"Percentage [%]\")\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:17:42.770349Z","iopub.execute_input":"2022-06-09T17:17:42.771185Z","iopub.status.idle":"2022-06-09T17:17:42.964308Z","shell.execute_reply.started":"2022-06-09T17:17:42.771151Z","shell.execute_reply":"2022-06-09T17:17:42.963297Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"It's clear that our dataset is unbalanced and this is a crucial point to keep in mind while modelling. 25% of customers had a default - it will be worth investigating these two groups separately to find some differences. First let's see how many unique customers do we have.","metadata":{}},{"cell_type":"code","source":"print(f'Number of unique customers: {train_raw[\"customer_ID\"].nunique()}')","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:17:42.965812Z","iopub.execute_input":"2022-06-09T17:17:42.966314Z","iopub.status.idle":"2022-06-09T17:17:43.973534Z","shell.execute_reply.started":"2022-06-09T17:17:42.966266Z","shell.execute_reply":"2022-06-09T17:17:43.972727Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.3 Customers presence </b>","metadata":{}},{"cell_type":"markdown","source":"We have about 459 000 unique customers.","metadata":{}},{"cell_type":"code","source":"cust_presence = train_raw.groupby(['customer_ID','target']).size().reset_index().rename(columns={0:'presence'})\ncust_presence.head()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:43.974625Z","iopub.execute_input":"2022-06-09T17:17:43.975351Z","iopub.status.idle":"2022-06-09T17:17:45.789563Z","shell.execute_reply.started":"2022-06-09T17:17:43.975316Z","shell.execute_reply":"2022-06-09T17:17:45.788314Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig, ax = plt.subplots(1,1, figsize=(15,5))\nsns.histplot(x='presence', data=cust_presence, hue='target', stat='percent', multiple=\"dodge\", bins=np.arange(0,14), ax=ax)\nax.bar_label(ax.containers[0], fmt='%.f%%')\nax.bar_label(ax.containers[1], fmt='%.f%%')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:45.791358Z","iopub.execute_input":"2022-06-09T17:17:45.791822Z","iopub.status.idle":"2022-06-09T17:17:46.322029Z","shell.execute_reply.started":"2022-06-09T17:17:45.791779Z","shell.execute_reply":"2022-06-09T17:17:46.320662Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The plot above shows how long are customer present in a dataset. It shows that 93% of them are visible for the entire full year. It's difficult from this graph to see how targets are distributed for the remaining customers, so let's zoom.","metadata":{}},{"cell_type":"code","source":"fig, ax = plt.subplots(1,1, figsize=(15,5))\nsns.histplot(x='presence', data=cust_presence, hue='target', stat='percent', multiple=\"dodge\", bins=np.arange(0,14), ax=ax)\nax.bar_label(ax.containers[0], fmt='%.2f%%')\nax.bar_label(ax.containers[1], fmt='%.2f%%')\nax.set_xlim(0,12)\nax.set_ylim(0,1)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:46.323472Z","iopub.execute_input":"2022-06-09T17:17:46.323779Z","iopub.status.idle":"2022-06-09T17:17:47.030862Z","shell.execute_reply.started":"2022-06-09T17:17:46.323749Z","shell.execute_reply":"2022-06-09T17:17:47.029724Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Now it's clear that customers who are for a short time in a database are more prone to churn. We can add now this information as an additional feature if we want.\nDuring a discussion [here](https://www.kaggle.com/competitions/amex-default-prediction/discussion/327597#1809185) there was a question who are customers with less than 13 observations. Let's investigate this.","metadata":{}},{"cell_type":"code","source":"short_customer_ids = list(cust_presence[cust_presence['presence']<13]['customer_ID'])\ngc.collect()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:47.032501Z","iopub.execute_input":"2022-06-09T17:17:47.032808Z","iopub.status.idle":"2022-06-09T17:17:47.187382Z","shell.execute_reply.started":"2022-06-09T17:17:47.032779Z","shell.execute_reply":"2022-06-09T17:17:47.186364Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"short_customers = train_raw[train_raw['customer_ID'].isin(short_customer_ids)][['customer_ID','S_2']]\nshort_customers['month'] = short_customers['S_2'].dt.month\nshort_customers['year'] = short_customers['S_2'].dt.year\nshort_customers.groupby(['year','month']).size()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:47.188909Z","iopub.execute_input":"2022-06-09T17:17:47.189258Z","iopub.status.idle":"2022-06-09T17:17:48.474174Z","shell.execute_reply.started":"2022-06-09T17:17:47.189224Z","shell.execute_reply":"2022-06-09T17:17:48.473230Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"len(short_customers)","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:48.475523Z","iopub.execute_input":"2022-06-09T17:17:48.475890Z","iopub.status.idle":"2022-06-09T17:17:48.484157Z","shell.execute_reply.started":"2022-06-09T17:17:48.475859Z","shell.execute_reply":"2022-06-09T17:17:48.483011Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"If all \"short\" customers are only there due to late entry then all with the same presence number would be in the same month.","metadata":{}},{"cell_type":"code","source":"n=2\nshort_customer_ids_n = list(cust_presence[cust_presence['presence']<=n]['customer_ID'])\n\nshort_customers = train_raw[train_raw['customer_ID'].isin(short_customer_ids_n)][['customer_ID','S_2']]\nshort_customers['month'] = short_customers['S_2'].dt.month\nshort_customers['year'] = short_customers['S_2'].dt.year\nprint(f\"Customers with presence equal to {n} observations.\")\nprint(short_customers.groupby(['year','month']).size())","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:48.485474Z","iopub.execute_input":"2022-06-09T17:17:48.485845Z","iopub.status.idle":"2022-06-09T17:17:49.378024Z","shell.execute_reply.started":"2022-06-09T17:17:48.485805Z","shell.execute_reply":"2022-06-09T17:17:49.376794Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"len(short_customers)","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:49.382292Z","iopub.execute_input":"2022-06-09T17:17:49.382680Z","iopub.status.idle":"2022-06-09T17:17:49.389156Z","shell.execute_reply.started":"2022-06-09T17:17:49.382646Z","shell.execute_reply":"2022-06-09T17:17:49.388031Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Now, it's clear that short customers are also these who dropped from the observation period as well. Let's see now data for an exemplary customer from this subsample.","metadata":{}},{"cell_type":"code","source":"cust_2obs_03_2017 = list(short_customers.loc[(short_customers['year']==2017)&(short_customers['month']==3),'customer_ID'])","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:49.390565Z","iopub.execute_input":"2022-06-09T17:17:49.390922Z","iopub.status.idle":"2022-06-09T17:17:49.403299Z","shell.execute_reply.started":"2022-06-09T17:17:49.390892Z","shell.execute_reply":"2022-06-09T17:17:49.402193Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sample_customer = np.random.choice(cust_2obs_03_2017)\ntrain_raw[train_raw[\"customer_ID\"]==sample_customer].head()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:49.404861Z","iopub.execute_input":"2022-06-09T17:17:49.406193Z","iopub.status.idle":"2022-06-09T17:17:50.424276Z","shell.execute_reply.started":"2022-06-09T17:17:49.406121Z","shell.execute_reply":"2022-06-09T17:17:50.423086Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.4 Correlations </b>\n\nBefore we dive in into distributions of the individual features let's check are there any highly correlated ones. Because it would take ags to calculate the entire correlation matrix I'll use an approximate method by using a sample of 25% of customers.","metadata":{}},{"cell_type":"code","source":"def sample_full_cust(df, cust_ratio):\n    n_customers = df['customer_ID'].nunique()\n    no_of_cust = int(n_customers*cust_ratio)\n    cust_ids = np.random.choice(df['customer_ID'].unique(), no_of_cust)\n    print(f'Number of customers sampled: {no_of_cust}')\n    ready_df = df[df['customer_ID'].isin(cust_ids)]\n    print(f'Number of rows sampled: {len(ready_df)} ({round(len(ready_df)/len(df)*100)}%)')\n    return ready_df","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:50.425873Z","iopub.execute_input":"2022-06-09T17:17:50.426234Z","iopub.status.idle":"2022-06-09T17:17:50.435130Z","shell.execute_reply.started":"2022-06-09T17:17:50.426205Z","shell.execute_reply":"2022-06-09T17:17:50.433480Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_samples = sample_full_cust(train_raw, 0.55)","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:50.436981Z","iopub.execute_input":"2022-06-09T17:17:50.437717Z","iopub.status.idle":"2022-06-09T17:17:54.582904Z","shell.execute_reply.started":"2022-06-09T17:17:50.437682Z","shell.execute_reply":"2022-06-09T17:17:54.581740Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"correlations = train_samples.corr().abs()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:17:54.584739Z","iopub.execute_input":"2022-06-09T17:17:54.585636Z","iopub.status.idle":"2022-06-09T17:20:45.973130Z","shell.execute_reply.started":"2022-06-09T17:17:54.585582Z","shell.execute_reply":"2022-06-09T17:20:45.971499Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"mask=np.triu(np.ones_like(correlations))\n\nfig, ax = plt.subplots(1,1, figsize=(16,12))\nsns.heatmap(correlations, ax=ax, mask=mask, cmap='YlOrBr')\nax.set_title(\"Correlation of features\")\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:20:45.975032Z","iopub.execute_input":"2022-06-09T17:20:45.975428Z","iopub.status.idle":"2022-06-09T17:20:47.639242Z","shell.execute_reply.started":"2022-06-09T17:20:45.975395Z","shell.execute_reply":"2022-06-09T17:20:47.637828Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The graph above shows us that most of features is not correlated but there are visible dark 'pixels' meaning we have some highly-correlated ones. In order to print them from the most correlated to the least we have to unstack the results.","metadata":{}},{"cell_type":"code","source":"unstacked = correlations.unstack()\nunstacked = unstacked.sort_values(ascending=False, kind=\"quicksort\").drop_duplicates().head(25)\nunstacked","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:20:47.641045Z","iopub.execute_input":"2022-06-09T17:20:47.641602Z","iopub.status.idle":"2022-06-09T17:20:47.665148Z","shell.execute_reply.started":"2022-06-09T17:20:47.641553Z","shell.execute_reply":"2022-06-09T17:20:47.664422Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Indeed we have a lot of highly correlated features. Worth keeping this in mind!\nLet's plot a correlated features to visualise them. Because at each run the sample is different results may vary a bit.","metadata":{}},{"cell_type":"code","source":"x1, y1 = unstacked.index[1]\nx2, y2 = unstacked.index[2]\nx3, y3 = unstacked.index[3]\n\nfig, ax = plt.subplots(1,3, figsize=(15,5))\nsns.scatterplot(x=x1, y=y1, data=train_samples, hue='target', ax=ax[0])\nsns.scatterplot(x=x2, y=y2, data=train_samples, hue='target', ax=ax[1])\nsns.scatterplot(x=x3, y=y3, data=train_samples, hue='target', ax=ax[2])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:20:47.666226Z","iopub.execute_input":"2022-06-09T17:20:47.666859Z","iopub.status.idle":"2022-06-09T17:23:33.110491Z","shell.execute_reply.started":"2022-06-09T17:20:47.666828Z","shell.execute_reply":"2022-06-09T17:23:33.109179Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"The plot at the beggining of this chapter may be a bit unclear if we are interested more in details. Therefore it would be better to look at the correlations on the level of each column group. As this matrix is symmetrical we can take only one side of it.","metadata":{}},{"cell_type":"code","source":"# code adapted from: https://www.kaggle.com/code/kellibelcher/amex-default-prediction-eda-lgbm-baseline\ncols_to_show = [c for c in train_samples.columns if (c.startswith('R'))]\ncorr=train_samples[cols_to_show].corr()\nmask=np.triu(np.ones_like(corr))[1:,:-1]\ncorr=corr.iloc[1:,:-1].copy()\n\nfig, ax = plt.subplots(figsize=(30,30))   \nsns.heatmap(corr, mask=mask, vmin=-1, vmax=1, center=0, annot=True, fmt='.2f', \n            cmap='coolwarm', annot_kws={'fontsize':10,'fontweight':'bold'}, cbar=False)\nax.tick_params(left=False,bottom=False)\nax.set_xticklabels(ax.get_xticklabels(), rotation=45, horizontalalignment='right',fontsize=12)\nax.set_yticklabels(ax.get_yticklabels(), fontsize=12)\nplt.title('Correlations between Risk Variables\\n', fontsize=16)\nplt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:39:32.998257Z","iopub.execute_input":"2022-06-09T17:39:32.998728Z","iopub.status.idle":"2022-06-09T17:39:40.373010Z","shell.execute_reply.started":"2022-06-09T17:39:32.998693Z","shell.execute_reply":"2022-06-09T17:39:40.371891Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# code adapted from: https://www.kaggle.com/code/kellibelcher/amex-default-prediction-eda-lgbm-baseline\ncols_to_show = [c for c in train_samples.columns if (c.startswith('S'))]\ncorr=train_samples[cols_to_show].corr()\nmask=np.triu(np.ones_like(corr))[1:,:-1]\ncorr=corr.iloc[1:,:-1].copy()\n\nfig, ax = plt.subplots(figsize=(15,15))   \nsns.heatmap(corr, mask=mask, vmin=-1, vmax=1, center=0, annot=True, fmt='.2f', \n            cmap='coolwarm', annot_kws={'fontsize':10,'fontweight':'bold'}, cbar=False)\nax.tick_params(left=False,bottom=False)\nax.set_xticklabels(ax.get_xticklabels(), rotation=45, horizontalalignment='right',fontsize=12)\nax.set_yticklabels(ax.get_yticklabels(), fontsize=12)\nplt.title('Correlations between Spend Variables\\n', fontsize=16)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:38:50.083462Z","iopub.execute_input":"2022-06-09T17:38:50.083970Z","iopub.status.idle":"2022-06-09T17:38:54.417892Z","shell.execute_reply.started":"2022-06-09T17:38:50.083918Z","shell.execute_reply":"2022-06-09T17:38:54.416394Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# <b> 3.5 Categorical features </b>\n\nWe are given by the organizers a list of categorical features. Let's take a quick look at them.","metadata":{}},{"cell_type":"code","source":"train_raw[categorical_features].sample(10)","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:23:40.906411Z","iopub.execute_input":"2022-06-09T17:23:40.906732Z","iopub.status.idle":"2022-06-09T17:23:41.354324Z","shell.execute_reply.started":"2022-06-09T17:23:40.906703Z","shell.execute_reply":"2022-06-09T17:23:41.353070Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"What are the unique values per categorical column?","metadata":{}},{"cell_type":"code","source":"for cf in categorical_features:\n    print(cf, list(train_raw[cf].unique()))","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:23:41.356242Z","iopub.execute_input":"2022-06-09T17:23:41.356821Z","iopub.status.idle":"2022-06-09T17:23:41.800429Z","shell.execute_reply.started":"2022-06-09T17:23:41.356767Z","shell.execute_reply":"2022-06-09T17:23:41.798854Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"How many missing values ber category?","metadata":{}},{"cell_type":"code","source":"train_raw[categorical_features].isna().sum().div(len(train_raw)).sort_values(ascending=False)","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:23:41.802119Z","iopub.execute_input":"2022-06-09T17:23:41.802516Z","iopub.status.idle":"2022-06-09T17:23:41.908757Z","shell.execute_reply.started":"2022-06-09T17:23:41.802482Z","shell.execute_reply":"2022-06-09T17:23:41.907501Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Distributions","metadata":{}},{"cell_type":"code","source":"def show_kdeplots(letter, figsize):   \n    cols = [c for c in train_samples.columns if (c.startswith((letter,'t'))) & (c not in categorical_features)]\n    df_tmp = train_samples[cols]\n    plt_cols = 5\n    plt_rows = math.ceil(len(cols)/plt_cols)\n\n    fig, axes = plt.subplots(plt_rows, plt_cols, figsize=figsize)\n    for i, ax in enumerate(axes.reshape(-1)):\n        if i<len(cols)-1:\n            sns.kdeplot(x=cols[i], hue='target', hue_order=[1,0], label=['Default','Paid'], data=df_tmp, \n                        fill=True, linewidth=2, legend=False, ax=ax)\n        ax.tick_params(left=False, bottom=False, labelsize=5)\n        ax.xaxis.get_label().set_fontsize(10)\n        ax.set_ylabel('')\n\n    sns.despine(bottom=True, trim=True)\n    plt.tight_layout(rect=[0, 0.2, 1, 0.99])\n    plt.show()","metadata":{"_kg_hide-input":true,"execution":{"iopub.status.busy":"2022-06-09T17:23:41.910370Z","iopub.execute_input":"2022-06-09T17:23:41.910745Z","iopub.status.idle":"2022-06-09T17:23:41.923528Z","shell.execute_reply.started":"2022-06-09T17:23:41.910715Z","shell.execute_reply":"2022-06-09T17:23:41.922090Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"show_kdeplots('D', (15,30))","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:23:41.925327Z","iopub.execute_input":"2022-06-09T17:23:41.925775Z","iopub.status.idle":"2022-06-09T17:34:49.929620Z","shell.execute_reply.started":"2022-06-09T17:23:41.925737Z","shell.execute_reply":"2022-06-09T17:34:49.928162Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Let's see now some distributions for some random numerical column.","metadata":{}},{"cell_type":"code","source":"show_kdeplots('P', (15,5))","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:34:49.932172Z","iopub.execute_input":"2022-06-09T17:34:49.932688Z","iopub.status.idle":"2022-06-09T17:35:17.599643Z","shell.execute_reply.started":"2022-06-09T17:34:49.932634Z","shell.execute_reply":"2022-06-09T17:35:17.598556Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig, ax = plt.subplots(1,1, figsize=(10,5))\nsns.kdeplot(data=train_raw, x='B_23', hue='target', palette=[\"#3385ff\", \"#cc0000\"], ax=ax)\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:42:20.428246Z","iopub.execute_input":"2022-06-09T17:42:20.428757Z","iopub.status.idle":"2022-06-09T17:42:42.265626Z","shell.execute_reply.started":"2022-06-09T17:42:20.428719Z","shell.execute_reply":"2022-06-09T17:42:42.264840Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"This column shows a huge skewness. To plot it better let's use a logarithmic scale or just zoom in.","metadata":{}},{"cell_type":"code","source":"fig, ax = plt.subplots(1,2, figsize=(15,5))\nsns.kdeplot(data=train_raw, x='B_23', hue='target', palette=[\"#3385ff\", \"#cc0000\"], ax=ax[0])\nax[0].set_xscale('log')\nsns.kdeplot(data=train_raw, x='B_23', hue='target', palette=[\"#3385ff\", \"#cc0000\"], ax=ax[1])\nax[1].set_xlim([0,0.1])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-06-09T17:42:42.267164Z","iopub.execute_input":"2022-06-09T17:42:42.267652Z","iopub.status.idle":"2022-06-09T17:43:26.493645Z","shell.execute_reply.started":"2022-06-09T17:42:42.267617Z","shell.execute_reply":"2022-06-09T17:43:26.492364Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"That's all for now folks! I will add something here from time to time but if you like this so far, don't hesitate to upvote. Thanks!\n\n<b> Happy Kaggling! </b>","metadata":{}}]}