{"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":7921029,"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## Load the data","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 \n\ndataPath = \"/kaggle/input/home-credit-credit-risk-model-stability/\"","metadata":{"execution":{"iopub.status.busy":"2024-06-18T01:24:27.722857Z","iopub.execute_input":"2024-06-18T01:24:27.723393Z","iopub.status.idle":"2024-06-18T01:24:32.010975Z","shell.execute_reply.started":"2024-06-18T01:24:27.723350Z","shell.execute_reply":"2024-06-18T01:24:32.009556Z"},"trusted":true},"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#         if col[-1] in (\"D\"):\n#             df = df.with_columns(pl.col(col).cast(pl.Date).alias(col))\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()\n            if not 'Unknown' in new_categories:\n                new_categories.append('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":"2024-06-18T02:07:32.298200Z","iopub.execute_input":"2024-06-18T02:07:32.298825Z","iopub.status.idle":"2024-06-18T02:07:32.311545Z","shell.execute_reply.started":"2024-06-18T02:07:32.298782Z","shell.execute_reply":"2024-06-18T02:07:32.309759Z"},"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_person_2 = pl.read_csv(dataPath + \"csv_files/train/train_person_2.csv\").pipe(set_table_dtypes)\ntrain_tax_1 = pl.read_csv(dataPath + \"csv_files/train/train_tax_registry_a_1.csv\").pipe(set_table_dtypes)\n# train_credit_bureau_a_2 = pl.concat(\n#     [pl.read_csv(dataPath + f\"csv_files/train/train_credit_bureau_a_2_{i}.csv\").pipe(set_table_dtypes) for i in range(11)],\n#     how=\"vertical_relaxed\",\n# )\ntrain_deposit_1 = pl.read_csv(dataPath + \"csv_files/train/train_deposit_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":"2024-06-18T02:07:39.173613Z","iopub.execute_input":"2024-06-18T02:07:39.174052Z","iopub.status.idle":"2024-06-18T02:08:02.961885Z","shell.execute_reply.started":"2024-06-18T02:07:39.174019Z","shell.execute_reply":"2024-06-18T02:08:02.957237Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def filter_cols(df): #Remove those with an average is_null exceeding 0.95 and those that do not fall within the range 1 < nunique < 200.\n        for col in df.columns:\n            if col not in [\"target\", \"case_id\", \"WEEK_NUM\"]:\n                isnull = df[col].is_null().mean()\n                if isnull > 0 and col[-1] == 'A':\n                    df = df.with_columns(df[col].fill_null(0))\n\n        for col in df.columns:\n            if (col not in [\"target\", \"case_id\", \"WEEK_NUM\", ]) & (df[col].dtype == pl.String):\n                freq = df[col].n_unique()\n\n                if freq == 1:\n                    df = df.drop(col)\n        return df","metadata":{"execution":{"iopub.status.busy":"2024-06-18T01:24:54.069567Z","iopub.execute_input":"2024-06-18T01:24:54.070033Z","iopub.status.idle":"2024-06-18T01:24:54.080289Z","shell.execute_reply.started":"2024-06-18T01:24:54.069994Z","shell.execute_reply":"2024-06-18T01:24:54.078590Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print(train_basetable.columns)\n# print(train_static_cb.columns)\n# print(train_static_cb[].is_null().sum() / len(train_static_cb))\n# print(len(train_static.columns))\n# train_static = train_static.pipe(filter_cols)\n# print(train_static.null_count().sum())\n# print(len(train_static.columns))\n# for a in train_static.columns:\n#     if a[-1] == 'A' or a[-1] == 'M':\n#         nanratio = train_static[a].is_null().mean()\n#         print(a, nanratio)\n# print(len(train_static_cb.columns))\n# train_static_cb = train_static_cb.pipe(filter_cols)\n# print(train_static_cb.null_count().sum())\n# for a in train_static_cb.columns:\n#     if a[-1] == 'A' or a[-1] == 'M':\n#         nanratio = train_static_cb[a].is_null().mean()\n#         print(a, nanratio)\n# print(nan_ratio)","metadata":{"execution":{"iopub.status.busy":"2024-06-18T01:24:54.082200Z","iopub.execute_input":"2024-06-18T01:24:54.083428Z","iopub.status.idle":"2024-06-18T01:24:57.012087Z","shell.execute_reply.started":"2024-06-18T01:24:54.083356Z","shell.execute_reply":"2024-06-18T01:24:57.010508Z"},"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_tax_1 = pl.read_csv(dataPath + \"csv_files/test/test_tax_registry_a_1.csv\").pipe(set_table_dtypes)\ntest_deposit_1 = pl.read_csv(dataPath + \"csv_files/test/test_deposit_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":"2024-06-18T02:23:30.905676Z","iopub.execute_input":"2024-06-18T02:23:30.906752Z","iopub.status.idle":"2024-06-18T02:23:30.994065Z","shell.execute_reply.started":"2024-06-18T02:23:30.906699Z","shell.execute_reply":"2024-06-18T02:23:30.992565Z"},"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)\nprint(train_person_1_feats_1.null_count())\n# mainoccupationinc_384A: 客户主要收入金额，此处取出最大值\n# incometype_1044T：客户收入类型，判断是否为个体经营\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# train_person_1_feats_2 = train_person_1_feats_2.fill_nan('unknown')\n# print(train_person_1_feats_2.null_count())\n# print(train_person_1_feats_2.head())\n# 抽取贷款人房屋类型信息\n\ntrain_person_1_feats_3 = train_person_1.group_by(\"case_id\").agg(\n    pl.first('empl_industry_691L').alias(f\"first_empl_industry_691L\"),\n    pl.first('empl_employedtotal_800L').alias(f\"first_empl_employedtotal_800L\"),\n    pl.first('familystate_447L').alias(f\"first_familystate_447L\"),\n)\n\n# train_person_2_feats = train_person_2.group_by(\"case_id\").agg(\n#     pl.first('empls_economicalst_849M').alias(f\"first_empls_economicalst_849M\"),\n#     pl.first('empls_employer_name_740M').alias(f\"first_empls_employer_name_740M\"),\n#     pl.first('familystate_447L').alias(f\"first_familystate_447L\"),\n# )\n\n# print(train_person_1_feats_3.head())\n# print(train_person_1_feats_3.null_count())\n# empl_industry_691L：行业信息\n\n# train_person_1_feats_4 = train_person_1.group_by(\"case_id\").agg(\n#     pl.first('empl_employedtotal_800L').alias(f\"first_empl_employedtotal_800L\")\n# )\n# print(train_person_1_feats_4.head())\n# # empl_employedtotal_800L就业时长信息\n\n# train_person_1_feats_5 = train_person_1.group_by(\"case_id\").agg(\n#     pl.first('familystate_447L').alias(f\"first_familystate_447L\")\n# )\n# print(train_person_1_feats_5.head())\n# # familystate_447L 家庭情况\n\n\ntrain_tax_a_feats_1 = train_tax_1.group_by(\"case_id\").agg(\n    pl.col('amount_4527230A').fill_null(strategy=\"zero\").map_elements(lambda x: x.max() - x.min()).alias(f\"max-min_gap_depth2_amount_4527230A\"),\n    pl.col('amount_4527230A').max().alias(f\"max_amount_4527230A\"),\n    pl.col('amount_4527230A').min().alias(f\"min_amount_4527230A\"),\n    pl.col('amount_4527230A').mean().alias(f\"mean_amount_4527230A\"),\n    pl.col('amount_4527230A').std().alias(f\"std_amount_4527230A\"),\n)\n# print(train_tax_a_feats_1.null_count())\n# 交税情况，取差值，反应收入变化\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# print(train_credit_bureau_b_2_feats.head())\n# print(train_credit_bureau_b_2_feats.null_count())\n# pmts_pmtsoverdue_635A：逾期的贷款，选择逾期贷款的最大金额\n# pmts_dpdvalue_108P：是否逾期超过1个月\n\ntrain_deposit_feats = train_deposit_1.group_by(\"case_id\").agg(\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").max().alias(\"max_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").min().alias(\"min_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").mean().alias(\"mean_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").std().alias(\"std_amount_416A\"),\n)\n# print(train_deposit_feats.null_count())\n# amount_416A: 存款数额\n\n# train_credit_bureau_a_2_feats_1 = train_credit_bureau_a_2.group_by(\"case_id\").agg(\n#     pl.col('collater_valueofguarantee_1124L').fill_null(strategy=\"zero\").map_elements(lambda x: x.max() - x.min()).alias(f\"max-min_gap_depth2_collater_valueofguarantee_1124L\"),\n#     pl.col('collater_valueofguarantee_1124L').max().alias(f\"max_collater_valueofguarantee_1124L\"),\n#     pl.col('collater_valueofguarantee_1124L').min().alias(f\"min_collater_valueofguarantee_1124L\")\n# )\n# print(train_credit_bureau_a_2_feats_1.head())\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_person_1_feats_3, how=\"left\", on=\"case_id\"\n).join(\n    train_credit_bureau_b_2_feats, how=\"left\", on=\"case_id\"\n).join(\n    train_tax_a_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    train_deposit_feats, how=\"left\", on=\"case_id\"\n)\n\n\nfor col in data.columns:\n    if (col[-1] in (\"D\",)):\n        data = data.with_columns(pl.col(col) - pl.col(\"date_decision\"))\n        data = data.with_columns(pl.col(col).dt.total_days())\n    if data[col].dtype == pl.String:\n        data = data.with_columns(data[col].fill_null('Unknown'))\n    elif data[col].dtype == pl.Boolean:\n        data = data.with_columns(data[col].fill_null(False))\n    else:\n        data = data.with_columns(data[col].fill_null(-9999))\n#     elif df[col].dtype.name in ['']\n#     print(data[col].dtype)\n\nprint(data.head())","metadata":{"execution":{"iopub.status.busy":"2024-06-18T02:18:03.721963Z","iopub.execute_input":"2024-06-18T02:18:03.722503Z","iopub.status.idle":"2024-06-18T02:18:14.622830Z","shell.execute_reply.started":"2024-06-18T02:18:03.722415Z","shell.execute_reply":"2024-06-18T02:18:14.621540Z"},"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_person_1_feats_3 = test_person_1.group_by(\"case_id\").agg(\n    pl.first('empl_industry_691L').alias(f\"first_empl_industry_691L\"),\n    pl.first('empl_employedtotal_800L').alias(f\"first_empl_employedtotal_800L\"),\n    pl.first('familystate_447L').alias(f\"first_familystate_447L\"),\n)\n\ntest_tax_a_feats_1 = test_tax_1.group_by(\"case_id\").agg(\n    pl.col('amount_4527230A').fill_null(strategy=\"zero\").map_elements(lambda x: x.max() - x.min()).alias(f\"max-min_gap_depth2_amount_4527230A\"),\n    pl.col('amount_4527230A').max().alias(f\"max_amount_4527230A\"),\n    pl.col('amount_4527230A').min().alias(f\"min_amount_4527230A\"),\n    pl.col('amount_4527230A').mean().alias(f\"mean_amount_4527230A\"),\n    pl.col('amount_4527230A').std().alias(f\"std_amount_4527230A\"),\n)\n\ntest_deposit_feats = test_deposit_1.group_by(\"case_id\").agg(\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").max().alias(\"max_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").min().alias(\"min_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").mean().alias(\"mean_amount_416A\"),\n    pl.col(\"amount_416A\").fill_null(strategy=\"zero\").std().alias(\"std_amount_416A\"),\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)\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_person_1_feats_3, how=\"left\", on=\"case_id\"\n).join(\n    test_tax_a_feats_1, how=\"left\", on=\"case_id\"\n).join(\n    test_deposit_feats, how=\"left\", on=\"case_id\"\n).join(\n    test_credit_bureau_b_2_feats, how=\"left\", on=\"case_id\"\n)\n\nfor col in data_submission.columns:\n    if (col[-1] in (\"D\",)):\n        data_submission = data_submission.with_columns(pl.col(col) - pl.col(\"date_decision\"))\n        data_submission = data_submission.with_columns(pl.col(col).dt.total_days())\n    if data_submission[col].dtype == pl.String:\n        data_submission = data_submission.with_columns(data_submission[col].fill_null('Unknown'))\n    elif data_submission[col].dtype == pl.Boolean:\n        data_submission = data_submission.with_columns(data_submission[col].fill_null(False))\n    else:\n        data_submission = data_submission.with_columns(data_submission[col].fill_null(-9999))\n#     elif df[col].dtype.name in ['']\n#     print(data[col].dtype)\n\nprint(data_submission.head())","metadata":{"execution":{"iopub.status.busy":"2024-06-18T02:43:11.035798Z","iopub.execute_input":"2024-06-18T02:43:11.038067Z","iopub.status.idle":"2024-06-18T02:43:11.114903Z","shell.execute_reply.started":"2024-06-18T02:43:11.037996Z","shell.execute_reply":"2024-06-18T02:43:11.112746Z"},"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)\nprint(X_train.head())\n\nfor df in [X_train, X_valid, X_test]:\n    df = convert_strings(df)","metadata":{"execution":{"iopub.status.busy":"2024-06-18T02:45:16.047355Z","iopub.execute_input":"2024-06-18T02:45:16.049051Z","iopub.status.idle":"2024-06-18T02:45:27.537787Z","shell.execute_reply.started":"2024-06-18T02:45:16.048996Z","shell.execute_reply":"2024-06-18T02:45:27.535789Z"},"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}\")\nprint(X_train.head())","metadata":{"execution":{"iopub.status.busy":"2024-06-18T02:18:39.063728Z","iopub.execute_input":"2024-06-18T02:18:39.064206Z","iopub.status.idle":"2024-06-18T02:18:39.088546Z","shell.execute_reply.started":"2024-06-18T02:18:39.064171Z","shell.execute_reply":"2024-06-18T02:18:39.086768Z"},"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":"2024-06-18T02:18:42.689022Z","iopub.execute_input":"2024-06-18T02:18:42.689500Z","iopub.status.idle":"2024-06-18T02:20:57.089675Z","shell.execute_reply.started":"2024-06-18T02:18:42.689447Z","shell.execute_reply":"2024-06-18T02:20:57.087844Z"},"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\n# print(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":"2024-06-18T02:21:01.014393Z","iopub.execute_input":"2024-06-18T02:21:01.014932Z","iopub.status.idle":"2024-06-18T02:21:34.218018Z","shell.execute_reply.started":"2024-06-18T02:21:01.014893Z","shell.execute_reply":"2024-06-18T02:21:34.216017Z"},"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":"2024-06-18T02:21:40.111808Z","iopub.execute_input":"2024-06-18T02:21:40.112284Z","iopub.status.idle":"2024-06-18T02:21:41.334108Z","shell.execute_reply.started":"2024-06-18T02:21:40.112245Z","shell.execute_reply":"2024-06-18T02:21:41.332571Z"},"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\nprint(X_submission.shape)\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":"2024-06-18T02:46:10.919073Z","iopub.execute_input":"2024-06-18T02:46:10.919575Z","iopub.status.idle":"2024-06-18T02:46:11.061286Z","shell.execute_reply.started":"2024-06-18T02:46:10.919535Z","shell.execute_reply":"2024-06-18T02:46:11.059652Z"},"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":"2024-06-18T02:46:16.864066Z","iopub.execute_input":"2024-06-18T02:46:16.864655Z","iopub.status.idle":"2024-06-18T02:46:16.876362Z","shell.execute_reply.started":"2024-06-18T02:46:16.864609Z","shell.execute_reply":"2024-06-18T02:46:16.874755Z"},"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":{}}]}