{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.13","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":7921029,"sourceType":"competition"}],"dockerImageVersionId":30646,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Home Credit Kaggle 2024 submission\n*Author : Sue Huynh - Date : February 2024*\n\nWelcome and thank you for visiting my work on the Home Credit 2024 competition!\n\n**About the 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 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\n\n\n**About my submission** - Inspired by the [Starter notebook](https://www.kaggle.com/code/jetakow/home-credit-2024-starter-notebook) provided by the host, I will\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 to save memory and clean train data\n* Use hyperparameters tuning (BayesSearchCV)\n* Train a LightGBM model","metadata":{}},{"cell_type":"markdown","source":"## Load libraries","metadata":{}},{"cell_type":"code","source":"import polars as pl\nimport polars.selectors as cs\nimport numpy as np\nimport pandas as pd\nfrom datetime import datetime\n\nimport lightgbm as lgb\nimport xgboost as xgb\nimport sklearn\nimport skopt\n\nfrom sklearn.model_selection import train_test_split, cross_val_score, StratifiedKFold\nfrom sklearn.metrics import roc_auc_score \nfrom skopt import BayesSearchCV\n\nimport matplotlib.pyplot as plt\n\ndataPath = \"/kaggle/input/home-credit-credit-risk-model-stability/\"","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2024-03-13T02:17:11.568319Z","iopub.execute_input":"2024-03-13T02:17:11.568759Z","iopub.status.idle":"2024-03-13T02:17:11.576784Z","shell.execute_reply.started":"2024-03-13T02:17:11.568730Z","shell.execute_reply":"2024-03-13T02:17:11.575301Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def to_pandas(df_data, cat_cols=None):\n    df_data = df_data.to_pandas()\n    \n    if cat_cols is None:\n        cat_cols = list(df_data.select_dtypes(\"object\").columns)\n    \n    df_data[cat_cols] = df_data[cat_cols].astype(\"category\")\n    \n    return df_data, cat_cols","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:17:11.584146Z","iopub.execute_input":"2024-03-13T02:17:11.584883Z","iopub.status.idle":"2024-03-13T02:17:11.595771Z","shell.execute_reply.started":"2024-03-13T02:17:11.584831Z","shell.execute_reply":"2024-03-13T02:17:11.594172Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def set_table_dtypes(df):\n    for col in df.columns:\n        if col in [\"case_id\", \"WEEK_NUM\", \"num_group1\", \"num_group2\"]:\n            df = df.with_columns(pl.col(col).cast(pl.Int64))\n        elif col in [\"date_decision\"]:\n            df = df.with_columns(pl.col(col).cast(pl.Date))\n        elif col[-1] in (\"P\", \"A\"):\n            df = df.with_columns(pl.col(col).cast(pl.Float64))\n        elif col[-1] in (\"M\",):\n            df = df.with_columns(pl.col(col).cast(pl.String))\n        elif col[-1] in (\"D\",):\n            df = df.with_columns(pl.col(col).cast(pl.Date))            \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":"2024-03-13T02:17:11.608067Z","iopub.execute_input":"2024-03-13T02:17:11.610444Z","iopub.status.idle":"2024-03-13T02:17:11.625845Z","shell.execute_reply.started":"2024-03-13T02:17:11.610394Z","shell.execute_reply":"2024-03-13T02:17:11.624452Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Load the data","metadata":{}},{"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_debitcard_1 = pl.read_csv(dataPath + \"csv_files/train/train_debitcard_1.csv\").pipe(set_table_dtypes)\ntrain_applprev_1_0 = pl.read_csv(dataPath + \"csv_files/train/train_applprev_1_0.csv\").pipe(set_table_dtypes)\ntrain_applprev_1_1 = pl.read_csv(dataPath + \"csv_files/train/train_applprev_1_1.csv\").pipe(set_table_dtypes)\ntrain_tax_registry_a_1 = pl.read_csv(dataPath + \"csv_files/train/train_tax_registry_a_1.csv\").pipe(set_table_dtypes)\ntrain_credit_bureau_a_1_0 = pl.read_csv(dataPath + \"csv_files/train/train_credit_bureau_a_1_0.csv\").pipe(set_table_dtypes)\ntrain_credit_bureau_a_1_1 = pl.read_csv(dataPath + \"csv_files/train/train_credit_bureau_a_1_1.csv\").pipe(set_table_dtypes)\ntrain_credit_bureau_a_1_2 = pl.read_csv(dataPath + \"csv_files/train/train_credit_bureau_a_1_2.csv\").pipe(set_table_dtypes)\ntrain_credit_bureau_a_1_3 = pl.read_csv(dataPath + \"csv_files/train/train_credit_bureau_a_1_3.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":"2024-03-13T02:17:11.628454Z","iopub.execute_input":"2024-03-13T02:17:11.628859Z","iopub.status.idle":"2024-03-13T02:18:59.609537Z","shell.execute_reply.started":"2024-03-13T02:17:11.628828Z","shell.execute_reply":"2024-03-13T02:18:59.607880Z"},"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_debitcard_1 = pl.read_csv(dataPath + \"csv_files/test/test_debitcard_1.csv\").pipe(set_table_dtypes)\ntest_applprev_1_0 = pl.read_csv(dataPath + \"csv_files/test/test_applprev_1_0.csv\").pipe(set_table_dtypes)\ntest_applprev_1_1 = pl.read_csv(dataPath + \"csv_files/test/test_applprev_1_1.csv\").pipe(set_table_dtypes)\ntest_tax_registry_a_1 = pl.read_csv(dataPath + \"csv_files/test/test_tax_registry_a_1.csv\").pipe(set_table_dtypes)\ntest_credit_bureau_a_1_0 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_a_1_0.csv\").pipe(set_table_dtypes)\ntest_credit_bureau_a_1_1 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_a_1_1.csv\").pipe(set_table_dtypes)\ntest_credit_bureau_a_1_2 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_a_1_2.csv\").pipe(set_table_dtypes)\ntest_credit_bureau_a_1_3 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_a_1_3.csv\").pipe(set_table_dtypes)\ntest_credit_bureau_a_1_4 = pl.read_csv(dataPath + \"csv_files/test/test_credit_bureau_a_1_4.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":"2024-03-13T02:18:59.612623Z","iopub.execute_input":"2024-03-13T02:18:59.613175Z","iopub.status.idle":"2024-03-13T02:18:59.745614Z","shell.execute_reply.started":"2024-03-13T02:18:59.613123Z","shell.execute_reply":"2024-03-13T02:18:59.744264Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Data Preprocessing\nCreate multiple aggregated features from long datasets - applprev, tax_registry, and credit_bureau. Here, I also replaced the original dataset with aggregate features to save memory for training later.","metadata":{}},{"cell_type":"markdown","source":"> appl_prev","metadata":{}},{"cell_type":"code","source":"train_applprev_1_0_feats_1 = train_applprev_1_0.group_by(\"case_id\").agg(\n    pl.col(\"outstandingdebt_522A\").max().alias(\"outstandingdebt_522A\"))\ntrain_applprev_1_1_feats_1 = train_applprev_1_1.group_by(\"case_id\").agg(\n    pl.col(\"outstandingdebt_522A\").max().alias(\"outstandingdebt_522A\"))\ntrain_applprev_1 = pl.concat([train_applprev_1_0_feats_1,train_applprev_1_1_feats_1])","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:18:59.749636Z","iopub.execute_input":"2024-03-13T02:18:59.750302Z","iopub.status.idle":"2024-03-13T02:19:02.609824Z","shell.execute_reply.started":"2024-03-13T02:18:59.750257Z","shell.execute_reply":"2024-03-13T02:19:02.608720Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_applprev_1_0_feats_1 = test_applprev_1_0.group_by(\"case_id\").agg(\n    pl.col(\"outstandingdebt_522A\").max().alias(\"outstandingdebt_522A\"))\ntest_applprev_1_1_feats_1 = test_applprev_1_1.group_by(\"case_id\").agg(\n    pl.col(\"outstandingdebt_522A\").max().alias(\"outstandingdebt_522A\"))\ntest_applprev_1 = pl.concat([test_applprev_1_0_feats_1,test_applprev_1_1_feats_1])","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:19:02.611494Z","iopub.execute_input":"2024-03-13T02:19:02.612271Z","iopub.status.idle":"2024-03-13T02:19:02.621381Z","shell.execute_reply.started":"2024-03-13T02:19:02.612226Z","shell.execute_reply":"2024-03-13T02:19:02.620141Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"> tax_registry","metadata":{}},{"cell_type":"code","source":"train_tax_registry_a_1_feats_1 = train_tax_registry_a_1.group_by(\"case_id\").agg(\n    pl.col(\"amount_4527230A\").max().alias(\"amount_4527230A\"))\ntrain_tax_registry_a_1_feats_2 = train_tax_registry_a_1.select([\"case_id\", \"amount_4527230A\", \"num_group1\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"amount_4527230A\": \"applied_amount_4527230A\"})\ntrain_tax_registry_a_1 = train_tax_registry_a_1_feats_1.join(train_tax_registry_a_1_feats_2, how = \"outer\", on = \"case_id\")","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:19:02.623401Z","iopub.execute_input":"2024-03-13T02:19:02.624139Z","iopub.status.idle":"2024-03-13T02:19:03.205595Z","shell.execute_reply.started":"2024-03-13T02:19:02.624096Z","shell.execute_reply":"2024-03-13T02:19:03.204509Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_tax_registry_a_1_feats_1 = test_tax_registry_a_1.group_by(\"case_id\").agg(\n    pl.col(\"amount_4527230A\").max().alias(\"amount_4527230A\"))\ntest_tax_registry_a_1_feats_2 = test_tax_registry_a_1.select([\"case_id\", \"amount_4527230A\", \"num_group1\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"amount_4527230A\": \"applied_amount_4527230A\"})\ntest_tax_registry_a_1 = test_tax_registry_a_1_feats_1.join(test_tax_registry_a_1_feats_2, how = \"outer\", on = \"case_id\")","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:19:03.211255Z","iopub.execute_input":"2024-03-13T02:19:03.211867Z","iopub.status.idle":"2024-03-13T02:19:03.225399Z","shell.execute_reply.started":"2024-03-13T02:19:03.211835Z","shell.execute_reply":"2024-03-13T02:19:03.224247Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"> credit_bureau","metadata":{}},{"cell_type":"code","source":"#train_credit_bureau_a_1_0\ntrain_credit_bureau_a_1_0_feats_1 = train_credit_bureau_a_1_0.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_train_credit_bureau_a_1_0 = []\nfor col in train_credit_bureau_a_1_0.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_train_credit_bureau_a_1_0.append(col)\ntrain_credit_bureau_a_1_0_cats = train_credit_bureau_a_1_0[selected_train_credit_bureau_a_1_0].unique()\n\ntrain_credit_bureau_a_1_0 = train_credit_bureau_a_1_0_cats.join(train_credit_bureau_a_1_0_feats_1, on = \"case_id\", how = \"left\")\n\n#train_credit_bureau_a_1_1\ntrain_credit_bureau_a_1_1_feats_1 = train_credit_bureau_a_1_1.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_train_credit_bureau_a_1_1 = []\nfor col in train_credit_bureau_a_1_1.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_train_credit_bureau_a_1_1.append(col)\ntrain_credit_bureau_a_1_1_cats = train_credit_bureau_a_1_1[selected_train_credit_bureau_a_1_1].unique()\n\ntrain_credit_bureau_a_1_1 = train_credit_bureau_a_1_1_cats.join(train_credit_bureau_a_1_1_feats_1, on = \"case_id\", how = \"left\")\n\n#train_credit_bureau_a_1_2\ntrain_credit_bureau_a_1_2_feats_1 = train_credit_bureau_a_1_2.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_train_credit_bureau_a_1_2 = []\nfor col in train_credit_bureau_a_1_2.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_train_credit_bureau_a_1_2.append(col)\ntrain_credit_bureau_a_1_2_cats = train_credit_bureau_a_1_2[selected_train_credit_bureau_a_1_2].unique()\n\ntrain_credit_bureau_a_1_2 = train_credit_bureau_a_1_2_cats.join(train_credit_bureau_a_1_2_feats_1, on = \"case_id\", how = \"left\")\n\n#train_credit_bureau_a_1_3\ntrain_credit_bureau_a_1_3_feats_1 = train_credit_bureau_a_1_3.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_train_credit_bureau_a_1_3 = []\nfor col in train_credit_bureau_a_1_3.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_train_credit_bureau_a_1_3.append(col)\ntrain_credit_bureau_a_1_3_cats = train_credit_bureau_a_1_3[selected_train_credit_bureau_a_1_3].unique()\n\ntrain_credit_bureau_a_1_3 = train_credit_bureau_a_1_3_cats.join(train_credit_bureau_a_1_3_feats_1, on = \"case_id\", how = \"left\")\n\ntrain_credit_bureau_a_1_feats = pl.concat([train_credit_bureau_a_1_0, \n                                           train_credit_bureau_a_1_1, train_credit_bureau_a_1_2, \n                                           train_credit_bureau_a_1_3], \n                                          how = \"vertical_relaxed\")","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:19:03.227155Z","iopub.execute_input":"2024-03-13T02:19:03.227599Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#test_credit_bureau_a_1_0\ntest_credit_bureau_a_1_0_feats_1 = test_credit_bureau_a_1_0.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_test_credit_bureau_a_1_0 = []\nfor col in test_credit_bureau_a_1_0.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_test_credit_bureau_a_1_0.append(col)\ntest_credit_bureau_a_1_0_cats = test_credit_bureau_a_1_0[selected_test_credit_bureau_a_1_0].unique()\n\ntest_credit_bureau_a_1_0_feats = test_credit_bureau_a_1_0_cats.join(test_credit_bureau_a_1_0_feats_1, on = \"case_id\", how = \"left\")\n\n#test_credit_bureau_a_1_1\ntest_credit_bureau_a_1_1_feats_1 = test_credit_bureau_a_1_1.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_test_credit_bureau_a_1_1 = []\nfor col in test_credit_bureau_a_1_1.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_test_credit_bureau_a_1_1.append(col)\ntest_credit_bureau_a_1_1_cats = test_credit_bureau_a_1_1[selected_test_credit_bureau_a_1_1].unique()\n\ntest_credit_bureau_a_1_1_feats = test_credit_bureau_a_1_1_cats.join(test_credit_bureau_a_1_1_feats_1, on = \"case_id\", how = \"left\")\n\n#test_credit_bureau_a_1_2\ntest_credit_bureau_a_1_2_feats_1 = test_credit_bureau_a_1_2.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_test_credit_bureau_a_1_2 = []\nfor col in test_credit_bureau_a_1_2.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_test_credit_bureau_a_1_2.append(col)\ntest_credit_bureau_a_1_2_cats = test_credit_bureau_a_1_2[selected_test_credit_bureau_a_1_2].unique()\n\ntest_credit_bureau_a_1_2_feats = test_credit_bureau_a_1_2_cats.join(test_credit_bureau_a_1_2_feats_1, on = \"case_id\", how = \"left\")\n\n#test_credit_bureau_a_1_3\ntest_credit_bureau_a_1_3_feats_1 = test_credit_bureau_a_1_3.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_test_credit_bureau_a_1_3 = []\nfor col in test_credit_bureau_a_1_3.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_test_credit_bureau_a_1_3.append(col)\ntest_credit_bureau_a_1_3_cats = test_credit_bureau_a_1_3[selected_test_credit_bureau_a_1_3].unique()\n\ntest_credit_bureau_a_1_3_feats = test_credit_bureau_a_1_3_cats.join(test_credit_bureau_a_1_3_feats_1, on = \"case_id\", how = \"left\")\n\n#test_credit_bureau_a_1_4\ntest_credit_bureau_a_1_4_feats_1 = test_credit_bureau_a_1_4.group_by(\"case_id\").agg(\n    pl.col(['totaloutstanddebtvalue_39A']).max().alias(\"totaloutstanddebtvalue_39A\"))\n\nselected_test_credit_bureau_a_1_4 = []\nfor col in test_credit_bureau_a_1_4.columns:\n    if col[-1] in (\"M\") or col == \"case_id\":\n        selected_test_credit_bureau_a_1_4.append(col)\ntest_credit_bureau_a_1_4_cats = test_credit_bureau_a_1_4[selected_test_credit_bureau_a_1_4].unique()\n\ntest_credit_bureau_a_1_4_feats = test_credit_bureau_a_1_4_cats.join(test_credit_bureau_a_1_4_feats_1, on = \"case_id\", how = \"left\")\n\ntest_credit_bureau_a_1_feats = pl.concat([test_credit_bureau_a_1_0, \n                                          test_credit_bureau_a_1_1, \n                                          test_credit_bureau_a_1_2, \n                                          test_credit_bureau_a_1_3, \n                                          test_credit_bureau_a_1_4], how = \"vertical_relaxed\")","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Feature Engineering","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.\n\n## debit card\ntrain_debitcard_1_feats_1 = train_debitcard_1.group_by(\"case_id\").agg(\n    pl.col(\"last180dayaveragebalance_704A\").max().alias(\"last180dayaveragebalance_704A\")\n)\n\n#Here num_group1=0 has special meaning, it is the person who applied for the loan.\ntrain_debitcard_1_feats_2 = train_debitcard_1.select([\"case_id\", \"num_group1\", \"last180dayaveragebalance_704A\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"last180dayaveragebalance_704A\": \"person_averagebalanceA\"})\n\n\n## person\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_selfemployedM\")\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_housetypeM\"})\n\n\n# Here we have num_group1 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\", \"P\"):\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\", \"P\"):\n        selected_static_cb_cols.append(col)\nprint(selected_static_cb_cols)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training preparation","metadata":{}},{"cell_type":"code","source":"# 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_debitcard_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_debitcard_1_feats_2, how=\"left\", on=\"case_id\"\n).join(\n    train_applprev_1_0_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_applprev_1_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_credit_bureau_a_1_feats, how=\"left\", on=\"case_id\"\n).join(\n    train_tax_registry_a_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_tax_registry_a_1_feats_2, 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":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_debitcard_1_feats_1 = test_debitcard_1.group_by(\"case_id\").agg(\n    pl.col(\"last180dayaveragebalance_704A\").max().alias(\"last180dayaveragebalance_704A\")\n)\n\ntest_debitcard_1_feats_2 = test_debitcard_1.select([\"case_id\", \"num_group1\", \"last180dayaveragebalance_704A\"]).filter(\n    pl.col(\"num_group1\") == 0\n).drop(\"num_group1\").rename({\"last180dayaveragebalance_704A\": \"person_averagebalanceA\"})\n\n\ntest_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_selfemployedM\")\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_housetypeM\"})\n\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)\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_debitcard_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_debitcard_1_feats_2, how=\"left\", on=\"case_id\"\n).join(\n    test_applprev_1_0_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_applprev_1_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_credit_bureau_a_1_feats, how=\"left\", on=\"case_id\"\n).join(\n    test_tax_registry_a_1_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_tax_registry_a_1_feats_2, 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":{"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)\n    datetime_cols = df.select_dtypes(include = ['datetime']).columns\n    df = df.drop(columns = datetime_cols, inplace = True)","metadata":{"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":"2024-03-13T02:17:03.229707Z","iopub.execute_input":"2024-03-13T02:17:03.230160Z","iopub.status.idle":"2024-03-13T02:17:03.276434Z","shell.execute_reply.started":"2024-03-13T02:17:03.230127Z","shell.execute_reply":"2024-03-13T02:17:03.274697Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Hyperparameter Tuning with BayesSearchCV","metadata":{}},{"cell_type":"code","source":"'''np.int = np.int64\nbayes_cv_tuner = BayesSearchCV(estimator = lgb.LGBMClassifier(boosting_type='gbdt', n_jobs=-1, verbose=2),\n        search_spaces = {\n        'max_depth': (0, 15),\n        'min_child_samples': (0, 5),\n        'max_bin': (100, 300),\n        'reg_lambda': (1, 5, 'log-uniform'),\n        'reg_alpha': (1, 5, 'log-uniform')\n        },\n        scoring = 'roc_auc', cv = StratifiedKFold(n_splits=2), n_iter = 30, verbose = 1, refit = True)\n'''","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''bayes_cv_tuner.fit(X_train, y_train)\nprint(\"best params: %s\" % str(bayes_cv_tuner.best_params_))\n'''","metadata":{"scrolled":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**Best Params**: \n> ('max_bin', 262), ('max_depth', 2), ('min_child_samples', 5), ('reg_alpha', 1), ('reg_lambda', 5)\n\n> ('max_bin', 291), ('max_depth', 2), ('min_child_samples', 0), ('reg_alpha', 5), ('reg_lambda', 3)\n","metadata":{}},{"cell_type":"code","source":"#BEST_PARAMS = dict((bayes_cv_tuner.best_params_))\nBEST_PARAMS = {'max_bin': 262,\n              'max_depth': 2,\n              'min_child_samples':5,\n              'reg_alpha': 1,\n              'reg_lamda': 5}","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:13:22.068700Z","iopub.status.idle":"2024-03-13T02:13:22.069291Z","shell.execute_reply.started":"2024-03-13T02:13:22.068986Z","shell.execute_reply":"2024-03-13T02:13:22.069009Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"FIXED_PARAMS={'boosting_type' : 'gbdt',\n              'objective': 'binary',\n              'metric': 'auc',\n              'is_unbalance': True,\n              'num_leaves' : 31,\n              'learning_rate' : 0.05,\n              'n_estimators': 1000,\n              'colsample_bytree': 0.7,\n              'colsample_bynode': 0.9,\n              'verbose': -1,\n              'random_state': 32,\n              'early_stopping_rounds': 100,\n              'extra_tree' : True}\nparams = {**FIXED_PARAMS, **BEST_PARAMS}\nparams","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:13:22.071284Z","iopub.status.idle":"2024-03-13T02:13:22.071695Z","shell.execute_reply.started":"2024-03-13T02:13:22.071500Z","shell.execute_reply":"2024-03-13T02:13:22.071516Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training LightGBM","metadata":{}},{"cell_type":"code","source":"#tuned_1\nlgb_train = lgb.Dataset(X_train, label=y_train)\nlgb_valid = lgb.Dataset(X_valid, label=y_valid, reference=lgb_train)\n\n\ngbm = lgb.train(\n    params,\n    lgb_train,\n    valid_sets=lgb_valid,\n    callbacks=[lgb.log_evaluation(100), lgb.early_stopping(100)]\n)","metadata":{"execution":{"iopub.status.busy":"2024-03-13T02:13:22.072848Z","iopub.status.idle":"2024-03-13T02:13:22.073541Z","shell.execute_reply.started":"2024-03-13T02:13:22.073274Z","shell.execute_reply":"2024-03-13T02:13:22.073300Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''\n#test\nlgb_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\": 5,\n    \"min_data_in_leaf\": 10,\n    \"num_leaves\": 31, \n    \"learning_rate\": 0.05,\n    \"max_bin\": 100, \n    \"n_estimators\": 1000,\n    \"colsample_bytree\": 0.7, \n    \"colsample_bynode\": 0.9,\n    \"verbose\": -1,\n    \"random_state\": 33,\n    \"reg_alpha\": 2.52, \n    \"reg_lambda\": 0.78, \n    \"extra_trees\":True\n}\n\ngbm = lgb.train(\n    params,\n    lgb_train,\n    valid_sets=lgb_valid,\n    callbacks=[lgb.log_evaluation(100), lgb.early_stopping(100)])\n'''","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''\n#reference from baseline models\nlgb_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\": 10,\n    \"learning_rate\": 0.05,\n    \"max_bin\": 255,\n    \"n_estimators\": 1000,\n    \"colsample_bytree\": 0.9, \n    \"colsample_bynode\": 0.9,\n    \"verbose\": -1,\n    \"random_state\": 42,\n    \"reg_alpha\": 0.1, \n    \"reg_lambda\": 3.25, \n    \"extra_trees\":True\n}\n\ngbm = lgb.train(\n    params,\n    lgb_train,\n    valid_sets=lgb_valid,\n    callbacks=[lgb.log_evaluation(100), lgb.early_stopping(100)]\n)'''","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"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":{"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":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Submission","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":{"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":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"submission","metadata":{"trusted":true},"execution_count":null,"outputs":[]}]}