{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":50160,"databundleVersionId":7602123,"sourceType":"competition"}],"dockerImageVersionId":30635,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Example Notebook\n\nWelcome to the example notebook for the Home Credit Kaggle competition. The goal of this competition is to determine how likely a customer is going to default on an issued loan. The main difference between the [first](https://www.kaggle.com/c/home-credit-default-risk) and this competition is that now your submission will be scored with a custom metric that will take into account how well the model performs in future. A decline in performance will be penalized. The goal is to create a model that is stable and performs well in the future.\n\nIn this notebook you will see how to:\n* Load the data\n* Join tables with Polars - a DataFrame library implemented in Rust language, designed to be blazingy fast and memory efficient.  \n* Create simple aggregation features\n* Train a LightGBM model\n* Create a submission table\n\n","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19"}},{"cell_type":"code","source":"import polars as pl\nimport numpy as np\nimport pandas as pd\nimport lightgbm as lgb\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import roc_auc_score \nimport time\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n\ndataPath = \"/kaggle/input/home-credit-credit-risk-model-stability/\"","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:05:19.572397Z","iopub.execute_input":"2024-02-22T13:05:19.57305Z","iopub.status.idle":"2024-02-22T13:05:19.747248Z","shell.execute_reply.started":"2024-02-22T13:05:19.57302Z","shell.execute_reply":"2024-02-22T13:05:19.745972Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Example","metadata":{}},{"cell_type":"code","source":"pd.set_option('display.max_columns', None) # we want to display all columns in this notebook\npd.set_option('display.max_rows', 500)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:57:35.157038Z","iopub.execute_input":"2024-02-22T12:57:35.157364Z","iopub.status.idle":"2024-02-22T12:57:35.161756Z","shell.execute_reply.started":"2024-02-22T12:57:35.157338Z","shell.execute_reply":"2024-02-22T12:57:35.160714Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"feature_def = pd.read_csv(dataPath + str('feature_definitions.csv'))\nfeature_def.head()","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:33:42.076397Z","iopub.execute_input":"2024-02-22T12:33:42.076737Z","iopub.status.idle":"2024-02-22T12:33:42.113177Z","shell.execute_reply.started":"2024-02-22T12:33:42.076709Z","shell.execute_reply":"2024-02-22T12:33:42.111929Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"feature_def.describe()","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:34:01.564633Z","iopub.execute_input":"2024-02-22T12:34:01.56498Z","iopub.status.idle":"2024-02-22T12:34:01.584393Z","shell.execute_reply.started":"2024-02-22T12:34:01.56495Z","shell.execute_reply":"2024-02-22T12:34:01.583532Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train = pd.read_csv(dataPath + str('csv_files/train/train_base.csv'))\ndf_test = pd.read_csv(dataPath + str('csv_files/test/test_base.csv'))\ndf_sample = pd.read_csv(dataPath + str('sample_submission.csv'))","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:39:11.451453Z","iopub.execute_input":"2024-02-22T12:39:11.451782Z","iopub.status.idle":"2024-02-22T12:39:11.912343Z","shell.execute_reply.started":"2024-02-22T12:39:11.451757Z","shell.execute_reply":"2024-02-22T12:39:11.911207Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train.head(10)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:39:18.981143Z","iopub.execute_input":"2024-02-22T12:39:18.98148Z","iopub.status.idle":"2024-02-22T12:39:18.992873Z","shell.execute_reply.started":"2024-02-22T12:39:18.981451Z","shell.execute_reply":"2024-02-22T12:39:18.991772Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_test.head(10)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:39:47.962697Z","iopub.execute_input":"2024-02-22T12:39:47.963029Z","iopub.status.idle":"2024-02-22T12:39:47.973512Z","shell.execute_reply.started":"2024-02-22T12:39:47.963004Z","shell.execute_reply":"2024-02-22T12:39:47.972453Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train.dtypes","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:41:00.758598Z","iopub.execute_input":"2024-02-22T12:41:00.758983Z","iopub.status.idle":"2024-02-22T12:41:00.767252Z","shell.execute_reply.started":"2024-02-22T12:41:00.758943Z","shell.execute_reply":"2024-02-22T12:41:00.765982Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train.MONTH = df_train.MONTH.astype('string')\ndf_test.MONTH = df_test.MONTH.astype('string')\n\ndf_train.date_decision = pd.to_datetime(df_train.date_decision)\ndf_test.date_decision = pd.to_datetime(df_test.date_decision)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:41:19.913679Z","iopub.execute_input":"2024-02-22T12:41:19.914055Z","iopub.status.idle":"2024-02-22T12:41:20.03606Z","shell.execute_reply.started":"2024-02-22T12:41:19.914025Z","shell.execute_reply":"2024-02-22T12:41:20.034922Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train.info()","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:41:43.822477Z","iopub.execute_input":"2024-02-22T12:41:43.822788Z","iopub.status.idle":"2024-02-22T12:41:43.879341Z","shell.execute_reply.started":"2024-02-22T12:41:43.822763Z","shell.execute_reply":"2024-02-22T12:41:43.878303Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_test.info()","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:43:25.094327Z","iopub.execute_input":"2024-02-22T12:43:25.094675Z","iopub.status.idle":"2024-02-22T12:43:25.106446Z","shell.execute_reply.started":"2024-02-22T12:43:25.094645Z","shell.execute_reply":"2024-02-22T12:43:25.105249Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df1 = pd.read_parquet(dataPath + str('parquet_files/train/train_static_0_0.parquet'))\ndf2 = pd.read_parquet(dataPath + str('parquet_files/train/train_static_0_1.parquet'))\nprint(df1.shape[0])\nprint(df2.shape[0])\ndf_train_static = pd.concat([df1,df2])\ndel(df1,df2)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:56:29.672789Z","iopub.execute_input":"2024-02-22T12:56:29.673144Z","iopub.status.idle":"2024-02-22T12:56:33.909625Z","shell.execute_reply.started":"2024-02-22T12:56:29.673118Z","shell.execute_reply":"2024-02-22T12:56:33.908829Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_static.head(10)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:57:41.530833Z","iopub.execute_input":"2024-02-22T12:57:41.531171Z","iopub.status.idle":"2024-02-22T12:57:41.674469Z","shell.execute_reply.started":"2024-02-22T12:57:41.531144Z","shell.execute_reply":"2024-02-22T12:57:41.673816Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_static.info(verbose=True, show_counts=True)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T12:59:26.430124Z","iopub.execute_input":"2024-02-22T12:59:26.430492Z","iopub.status.idle":"2024-02-22T12:59:27.98815Z","shell.execute_reply.started":"2024-02-22T12:59:26.430462Z","shell.execute_reply":"2024-02-22T12:59:27.987115Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_static_x = pd.read_parquet(dataPath + str('parquet_files/train/train_static_cb_0.parquet'))\ndf_train_static_x.head(10)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:00:11.821438Z","iopub.execute_input":"2024-02-22T13:00:11.822073Z","iopub.status.idle":"2024-02-22T13:00:12.88373Z","shell.execute_reply.started":"2024-02-22T13:00:11.82204Z","shell.execute_reply":"2024-02-22T13:00:12.88219Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_static_x.info(verbose=True, show_counts=True)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:00:40.295842Z","iopub.execute_input":"2024-02-22T13:00:40.296197Z","iopub.status.idle":"2024-02-22T13:00:40.888655Z","shell.execute_reply.started":"2024-02-22T13:00:40.296169Z","shell.execute_reply":"2024-02-22T13:00:40.887337Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## EDA","metadata":{}},{"cell_type":"code","source":"df_train.describe(include='all')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:01:21.588084Z","iopub.execute_input":"2024-02-22T13:01:21.588426Z","iopub.status.idle":"2024-02-22T13:01:21.808697Z","shell.execute_reply.started":"2024-02-22T13:01:21.588397Z","shell.execute_reply":"2024-02-22T13:01:21.807849Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(12,6))\nsns.countplot(x=df_train.MONTH)\nplt.xticks(rotation=90)\nplt.title('MONTH COUNT')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:06:37.117163Z","iopub.execute_input":"2024-02-22T13:06:37.11769Z","iopub.status.idle":"2024-02-22T13:06:38.22962Z","shell.execute_reply.started":"2024-02-22T13:06:37.11766Z","shell.execute_reply":"2024-02-22T13:06:38.228372Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(14,5))\nsns.countplot(x=df_train.WEEK_NUM)\nplt.xticks(rotation=90)\nplt.title('WEEK COUNT')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:08:12.615047Z","iopub.execute_input":"2024-02-22T13:08:12.615821Z","iopub.status.idle":"2024-02-22T13:08:13.487908Z","shell.execute_reply.started":"2024-02-22T13:08:12.615766Z","shell.execute_reply":"2024-02-22T13:08:13.486991Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#target val distribution\nplt.figure(figsize=(10,6))\nsns.countplot(x=df_train.target)\nplt.xticks(rotation=90)\nplt.title('TARGET DISTRIBUTION')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:09:29.084351Z","iopub.execute_input":"2024-02-22T13:09:29.086848Z","iopub.status.idle":"2024-02-22T13:09:29.373583Z","shell.execute_reply.started":"2024-02-22T13:09:29.086795Z","shell.execute_reply":"2024-02-22T13:09:29.37264Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"target_by_week = df_train.groupby(by=['WEEK_NUM'], observed=True)['target'].mean()\n\nplt.figure(figsize=(12,6))\nplt.bar(x=target_by_week.index, height=target_by_week, color='red')\nplt.title('TARGET MEAN BY WEEK')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:13:49.262705Z","iopub.execute_input":"2024-02-22T13:13:49.26306Z","iopub.status.idle":"2024-02-22T13:13:49.602113Z","shell.execute_reply.started":"2024-02-22T13:13:49.263033Z","shell.execute_reply":"2024-02-22T13:13:49.600686Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"target_by_month = df_train.groupby(by=['MONTH'], observed=True)['target'].mean()\n\nplt.figure(figsize=(14,3))\nplt.bar(x=target_by_month.index, height=target_by_month,\n        color='red')\nplt.xticks(rotation=90)\nplt.title('Target mean by month')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:15:39.131581Z","iopub.execute_input":"2024-02-22T13:15:39.132285Z","iopub.status.idle":"2024-02-22T13:15:39.616989Z","shell.execute_reply.started":"2024-02-22T13:15:39.132253Z","shell.execute_reply":"2024-02-22T13:15:39.615754Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train_static.describe(include='all')","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:24:42.577846Z","iopub.execute_input":"2024-02-22T13:24:42.578199Z","iopub.status.idle":"2024-02-22T13:24:51.46609Z","shell.execute_reply.started":"2024-02-22T13:24:42.578172Z","shell.execute_reply":"2024-02-22T13:24:51.464944Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"features_num = list(df_train_static.select_dtypes('number'))\nfeatures_num.remove('case_id')\nprint(features_num)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:25:04.479982Z","iopub.execute_input":"2024-02-22T13:25:04.480561Z","iopub.status.idle":"2024-02-22T13:25:05.215313Z","shell.execute_reply.started":"2024-02-22T13:25:04.48053Z","shell.execute_reply":"2024-02-22T13:25:05.213918Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"features_cat = df_train_static.columns.tolist()\nfeatures_cat = [el for el in features_cat if el not in features_num]\nfeatures_cat.remove('case_id')\nprint(features_cat)","metadata":{"execution":{"iopub.status.busy":"2024-02-22T13:25:46.548329Z","iopub.execute_input":"2024-02-22T13:25:46.548678Z","iopub.status.idle":"2024-02-22T13:25:46.555251Z","shell.execute_reply.started":"2024-02-22T13:25:46.548649Z","shell.execute_reply":"2024-02-22T13:25:46.55413Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"for f in features_num:\n    plt.figure(figsize=(8,4))\n    plt.subplot(1,2,1)\n    plt.hist()","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def set_table_dtypes(df: pl.DataFrame) -> pl.DataFrame:\n    # implement here all desired dtypes for tables\n    # the following is just an example\n    for col in df.columns:\n        # last letter of column name will help you determine the type\n        if col[-1] in (\"P\", \"A\"):\n            df = df.with_columns(pl.col(col).cast(pl.Float64).alias(col))\n\n    return df\n\ndef convert_strings(df: pd.DataFrame) -> pd.DataFrame:\n    for col in df.columns:  \n        if df[col].dtype.name in ['object', 'string']:\n            df[col] = df[col].astype(\"string\").astype('category')\n            current_categories = df[col].cat.categories\n            new_categories = current_categories.to_list() + [\"Unknown\"]\n            new_dtype = pd.CategoricalDtype(categories=new_categories, ordered=True)\n            df[col] = df[col].astype(new_dtype)\n    return df","metadata":{"execution":{"iopub.status.busy":"2024-02-22T10:52:05.464446Z","iopub.execute_input":"2024-02-22T10:52:05.465315Z","iopub.status.idle":"2024-02-22T10:52:05.47625Z","shell.execute_reply.started":"2024-02-22T10:52:05.465266Z","shell.execute_reply":"2024-02-22T10:52:05.474821Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_basetable = pl.read_csv(dataPath + \"csv_files/train/train_base.csv\")\ntrain_static = pl.concat(\n    [\n        pl.read_csv(dataPath + \"csv_files/train/train_static_0_0.csv\").pipe(set_table_dtypes),\n        pl.read_csv(dataPath + \"csv_files/train/train_static_0_1.csv\").pipe(set_table_dtypes),\n    ],\n    how=\"vertical_relaxed\",\n)\ntrain_static_cb = pl.read_csv(dataPath + \"csv_files/train/train_static_cb_0.csv\").pipe(set_table_dtypes)\ntrain_person_1 = pl.read_csv(dataPath + \"csv_files/train/train_person_1.csv\").pipe(set_table_dtypes) \ntrain_credit_bureau_b_2 = pl.read_csv(dataPath + \"csv_files/train/train_credit_bureau_b_2.csv\").pipe(set_table_dtypes) ","metadata":{"execution":{"iopub.status.busy":"2024-02-22T10:52:48.828336Z","iopub.execute_input":"2024-02-22T10:52:48.828914Z","iopub.status.idle":"2024-02-22T10:53:10.671889Z","shell.execute_reply.started":"2024-02-22T10:52:48.828875Z","shell.execute_reply":"2024-02-22T10:53:10.668146Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_basetable = pl.read_csv(dataPath + \"csv_files/test/test_base.csv\")\ntest_static = pl.concat(\n    [\n        pl.read_csv(dataPath + \"csv_files/test/test_static_0_0.csv\").pipe(set_table_dtypes),\n        pl.read_csv(dataPath + \"csv_files/test/test_static_0_1.csv\").pipe(set_table_dtypes),\n        pl.read_csv(dataPath + \"csv_files/test/test_static_0_2.csv\").pipe(set_table_dtypes),\n    ],\n    how=\"vertical_relaxed\",\n)\ntest_static_cb = pl.read_csv(dataPath + \"csv_files/test/test_static_cb_0.csv\").pipe(set_table_dtypes)\ntest_person_1 = pl.read_csv(dataPath + \"csv_files/test/test_person_1.csv\").pipe(set_table_dtypes) \ntest_credit_bureau_b_2 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_b_2.csv\").pipe(set_table_dtypes) ","metadata":{"execution":{"iopub.status.busy":"2024-02-22T10:53:10.67996Z","iopub.execute_input":"2024-02-22T10:53:10.681506Z","iopub.status.idle":"2024-02-22T10:53:10.840984Z","shell.execute_reply.started":"2024-02-22T10:53:10.681277Z","shell.execute_reply":"2024-02-22T10:53:10.838213Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_person_1.head()","metadata":{"execution":{"iopub.status.busy":"2024-02-22T10:53:10.844795Z","iopub.execute_input":"2024-02-22T10:53:10.845602Z","iopub.status.idle":"2024-02-22T10:53:10.897713Z","shell.execute_reply.started":"2024-02-22T10:53:10.845506Z","shell.execute_reply":"2024-02-22T10:53:10.894214Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Feature engineering\n\nIn this part, we can see a simple example of joining tables via `case_id`. Here the loading and joining is done with polars library. Polars library is blazingly fast and has much smaller memory footprint than pandas. ","metadata":{}},{"cell_type":"code","source":"# We need to use aggregation functions in tables with depth > 1, so tables that contain num_group1 column or \n# also num_group2 column.\ntrain_person_1_feats_1 = train_person_1.group_by(\"case_id\").agg(\n    pl.col(\"mainoccupationinc_384A\").max().alias(\"mainoccupationinc_384A_max\"),\n    (pl.col(\"incometype_1044T\") == \"SELFEMPLOYED\").max().alias(\"mainoccupationinc_384A_any_selfemployed\")\n)\n\n# Here num_group1=0 has special meaning, it is the person who applied for the loan.\ntrain_person_1_feats_2 = train_person_1.select([\"case_id\", \"num_group1\", \"housetype_905L\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"housetype_905L\": \"person_housetype\"})\n\n# Here we have num_goup1 and num_group2, so we need to aggregate again.\ntrain_credit_bureau_b_2_feats = train_credit_bureau_b_2.group_by(\"case_id\").agg(\n    pl.col(\"pmts_pmtsoverdue_635A\").max().alias(\"pmts_pmtsoverdue_635A_max\"),\n    (pl.col(\"pmts_dpdvalue_108P\") > 31).max().alias(\"pmts_dpdvalue_108P_over31\")\n)\n\n# We will process in this examples only A-type and M-type columns, so we need to select them.\nselected_static_cols = []\nfor col in train_static.columns:\n    if col[-1] in (\"A\", \"M\"):\n        selected_static_cols.append(col)\nprint(selected_static_cols)\n\nselected_static_cb_cols = []\nfor col in train_static_cb.columns:\n    if col[-1] in (\"A\", \"M\"):\n        selected_static_cb_cols.append(col)\nprint(selected_static_cb_cols)\n\n# Join all tables together.\ndata = train_basetable.join(\n    train_static.select([\"case_id\"]+selected_static_cols), how=\"left\", on=\"case_id\"\n).join(\n    train_static_cb.select([\"case_id\"]+selected_static_cb_cols), how=\"left\", on=\"case_id\"\n).join(\n    train_person_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_person_1_feats_2, how=\"left\", on=\"case_id\"\n).join(\n    train_credit_bureau_b_2_feats, how=\"left\", on=\"case_id\"\n)","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:26:09.582759Z","iopub.execute_input":"2024-02-21T11:26:09.583347Z","iopub.status.idle":"2024-02-21T11:26:11.382589Z","shell.execute_reply.started":"2024-02-21T11:26:09.583294Z","shell.execute_reply":"2024-02-21T11:26:11.38171Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_person_1_feats_1 = test_person_1.group_by(\"case_id\").agg(\n    pl.col(\"mainoccupationinc_384A\").max().alias(\"mainoccupationinc_384A_max\"),\n    (pl.col(\"incometype_1044T\") == \"SELFEMPLOYED\").max().alias(\"mainoccupationinc_384A_any_selfemployed\")\n)\n\ntest_person_1_feats_2 = test_person_1.select([\"case_id\", \"num_group1\", \"housetype_905L\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"housetype_905L\": \"person_housetype\"})\n\ntest_credit_bureau_b_2_feats = test_credit_bureau_b_2.group_by(\"case_id\").agg(\n    pl.col(\"pmts_pmtsoverdue_635A\").max().alias(\"pmts_pmtsoverdue_635A_max\"),\n    (pl.col(\"pmts_dpdvalue_108P\") > 31).max().alias(\"pmts_dpdvalue_108P_over31\")\n)\n\ndata_submission = test_basetable.join(\n    test_static.select([\"case_id\"]+selected_static_cols), how=\"left\", on=\"case_id\"\n).join(\n    test_static_cb.select([\"case_id\"]+selected_static_cb_cols), how=\"left\", on=\"case_id\"\n).join(\n    test_person_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_person_1_feats_2, how=\"left\", on=\"case_id\"\n).join(\n    test_credit_bureau_b_2_feats, how=\"left\", on=\"case_id\"\n)","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:26:12.72693Z","iopub.execute_input":"2024-02-21T11:26:12.727402Z","iopub.status.idle":"2024-02-21T11:26:12.744447Z","shell.execute_reply.started":"2024-02-21T11:26:12.727364Z","shell.execute_reply":"2024-02-21T11:26:12.743209Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"case_ids = data[\"case_id\"].unique().shuffle(seed=1)\ncase_ids_train, case_ids_test = train_test_split(case_ids, train_size=0.6, random_state=1)\ncase_ids_valid, case_ids_test = train_test_split(case_ids_test, train_size=0.5, random_state=1)\n\ncols_pred = []\nfor col in data.columns:\n    if col[-1].isupper() and col[:-1].islower():\n        cols_pred.append(col)\n\nprint(cols_pred)\n\ndef from_polars_to_pandas(case_ids: pl.DataFrame) -> pl.DataFrame:\n    return (\n        data.filter(pl.col(\"case_id\").is_in(case_ids))[[\"case_id\", \"WEEK_NUM\", \"target\"]].to_pandas(),\n        data.filter(pl.col(\"case_id\").is_in(case_ids))[cols_pred].to_pandas(),\n        data.filter(pl.col(\"case_id\").is_in(case_ids))[\"target\"].to_pandas()\n    )\n\nbase_train, X_train, y_train = from_polars_to_pandas(case_ids_train)\nbase_valid, X_valid, y_valid = from_polars_to_pandas(case_ids_valid)\nbase_test, X_test, y_test = from_polars_to_pandas(case_ids_test)\n\nfor df in [X_train, X_valid, X_test]:\n    df = convert_strings(df)","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:26:19.466384Z","iopub.execute_input":"2024-02-21T11:26:19.466805Z","iopub.status.idle":"2024-02-21T11:26:29.084141Z","shell.execute_reply.started":"2024-02-21T11:26:19.466765Z","shell.execute_reply":"2024-02-21T11:26:29.082364Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(f\"Train: {X_train.shape}\")\nprint(f\"Valid: {X_valid.shape}\")\nprint(f\"Test: {X_test.shape}\")","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:26:29.086236Z","iopub.execute_input":"2024-02-21T11:26:29.08663Z","iopub.status.idle":"2024-02-21T11:26:29.093734Z","shell.execute_reply.started":"2024-02-21T11:26:29.086593Z","shell.execute_reply":"2024-02-21T11:26:29.092252Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training LightGBM\n\nMinimal example of LightGBM training is shown below.","metadata":{}},{"cell_type":"code","source":"lgb_train = lgb.Dataset(X_train, label=y_train)\nlgb_valid = lgb.Dataset(X_valid, label=y_valid, reference=lgb_train)\n\nparams = {\n    \"boosting_type\": \"gbdt\",\n    \"objective\": \"binary\",\n    \"metric\": \"auc\",\n    \"max_depth\": 3,\n    \"num_leaves\": 31,\n    \"learning_rate\": 0.05,\n    \"feature_fraction\": 0.9,\n    \"bagging_fraction\": 0.8,\n    \"bagging_freq\": 5,\n    \"n_estimators\": 1000,\n    \"verbose\": -1,\n}\n\ngbm = lgb.train(\n    params,\n    lgb_train,\n    valid_sets=lgb_valid,\n    callbacks=[lgb.log_evaluation(50), lgb.early_stopping(10)]\n)","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:26:46.753042Z","iopub.execute_input":"2024-02-21T11:26:46.753504Z","iopub.status.idle":"2024-02-21T11:28:11.072442Z","shell.execute_reply.started":"2024-02-21T11:26:46.753468Z","shell.execute_reply":"2024-02-21T11:28:11.071363Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Evaluation with AUC and then comparison with the stability metric is shown below.","metadata":{}},{"cell_type":"code","source":"for base, X in [(base_train, X_train), (base_valid, X_valid), (base_test, X_test)]:\n    y_pred = gbm.predict(X, num_iteration=gbm.best_iteration)\n    base[\"score\"] = y_pred\n\nprint(f'The AUC score on the train set is: {roc_auc_score(base_train[\"target\"], base_train[\"score\"])}') \nprint(f'The AUC score on the valid set is: {roc_auc_score(base_valid[\"target\"], base_valid[\"score\"])}') \nprint(f'The AUC score on the test set is: {roc_auc_score(base_test[\"target\"], base_test[\"score\"])}')  ","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:28:33.05005Z","iopub.execute_input":"2024-02-21T11:28:33.050496Z","iopub.status.idle":"2024-02-21T11:28:56.27111Z","shell.execute_reply.started":"2024-02-21T11:28:33.050462Z","shell.execute_reply":"2024-02-21T11:28:56.269736Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def gini_stability(base, w_fallingrate=88.0, w_resstd=-0.5):\n    gini_in_time = base.loc[:, [\"WEEK_NUM\", \"target\", \"score\"]]\\\n        .sort_values(\"WEEK_NUM\")\\\n        .groupby(\"WEEK_NUM\")[[\"target\", \"score\"]]\\\n        .apply(lambda x: 2*roc_auc_score(x[\"target\"], x[\"score\"])-1).tolist()\n    \n    x = np.arange(len(gini_in_time))\n    y = gini_in_time\n    a, b = np.polyfit(x, y, 1)\n    y_hat = a*x + b\n    residuals = y - y_hat\n    res_std = np.std(residuals)\n    avg_gini = np.mean(gini_in_time)\n    return avg_gini + w_fallingrate * min(0, a) + w_resstd * res_std\n\nstability_score_train = gini_stability(base_train)\nstability_score_valid = gini_stability(base_valid)\nstability_score_test = gini_stability(base_test)\n\nprint(f'The stability score on the train set is: {stability_score_train}') \nprint(f'The stability score on the valid set is: {stability_score_valid}') \nprint(f'The stability score on the test set is: {stability_score_test}')  ","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:29:11.470254Z","iopub.execute_input":"2024-02-21T11:29:11.470673Z","iopub.status.idle":"2024-02-21T11:29:12.645586Z","shell.execute_reply.started":"2024-02-21T11:29:11.47064Z","shell.execute_reply":"2024-02-21T11:29:12.644402Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Submission\n\nScoring the submission dataset is below, we need to take care of new categories. Then we save the score as a last step. ","metadata":{}},{"cell_type":"code","source":"X_submission = data_submission[cols_pred].to_pandas()\nX_submission = convert_strings(X_submission)\ncategorical_cols = X_train.select_dtypes(include=['category']).columns\n\nfor col in categorical_cols:\n    train_categories = set(X_train[col].cat.categories)\n    submission_categories = set(X_submission[col].cat.categories)\n    new_categories = submission_categories - train_categories\n    X_submission.loc[X_submission[col].isin(new_categories), col] = \"Unknown\"\n    new_dtype = pd.CategoricalDtype(categories=train_categories, ordered=True)\n    X_train[col] = X_train[col].astype(new_dtype)\n    X_submission[col] = X_submission[col].astype(new_dtype)\n\ny_submission_pred = gbm.predict(X_submission, num_iteration=gbm.best_iteration)","metadata":{"execution":{"iopub.status.busy":"2024-02-21T11:30:00.507114Z","iopub.execute_input":"2024-02-21T11:30:00.507603Z","iopub.status.idle":"2024-02-21T11:30:00.644296Z","shell.execute_reply.started":"2024-02-21T11:30:00.507564Z","shell.execute_reply":"2024-02-21T11:30:00.64265Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"submission = pd.DataFrame({\n    \"case_id\": data_submission[\"case_id\"].to_numpy(),\n    \"score\": y_submission_pred\n}).set_index('case_id')\nsubmission.to_csv(\"./submission.csv\")","metadata":{"execution":{"iopub.status.busy":"2024-02-07T21:27:23.947771Z","iopub.execute_input":"2024-02-07T21:27:23.948164Z","iopub.status.idle":"2024-02-07T21:27:23.96104Z","shell.execute_reply.started":"2024-02-07T21:27:23.948128Z","shell.execute_reply":"2024-02-07T21:27:23.959969Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Best of luck, and most importantly, enjoy the process of learning and discovery! \n\n<img src=\"https://i.imgur.com/obVWIBh.png\" alt=\"Image\" width=\"700\"/>","metadata":{}}]}