{"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":"code","source":"import polars as pl\nimport numpy as np\nimport pandas as pd\nimport lightgbm as lgb\nimport xgboost as xgb\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":"2024-02-06T04:42:32.4461Z","iopub.execute_input":"2024-02-06T04:42:32.447156Z","iopub.status.idle":"2024-02-06T04:42:37.140933Z","shell.execute_reply.started":"2024-02-06T04:42:32.447087Z","shell.execute_reply":"2024-02-06T04:42:37.139637Z"},"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\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":"2024-02-06T04:42:37.143675Z","iopub.execute_input":"2024-02-06T04:42:37.144047Z","iopub.status.idle":"2024-02-06T04:42:37.155559Z","shell.execute_reply.started":"2024-02-06T04:42:37.144017Z","shell.execute_reply":"2024-02-06T04:42:37.153855Z"},"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":"2024-02-06T04:42:37.157982Z","iopub.execute_input":"2024-02-06T04:42:37.158496Z","iopub.status.idle":"2024-02-06T04:42:57.239525Z","shell.execute_reply.started":"2024-02-06T04:42:37.158448Z","shell.execute_reply":"2024-02-06T04:42:57.238571Z"},"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":"2024-02-06T04:42:57.24063Z","iopub.execute_input":"2024-02-06T04:42:57.240939Z","iopub.status.idle":"2024-02-06T04:42:57.311912Z","shell.execute_reply.started":"2024-02-06T04:42:57.24091Z","shell.execute_reply":"2024-02-06T04:42:57.309889Z"},"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":"2024-02-06T04:42:57.315566Z","iopub.execute_input":"2024-02-06T04:42:57.316147Z","iopub.status.idle":"2024-02-06T04:42:59.034123Z","shell.execute_reply.started":"2024-02-06T04:42:57.316111Z","shell.execute_reply":"2024-02-06T04:42:59.033242Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_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\ntest_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\ntest_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\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":"2024-02-06T04:42:59.035308Z","iopub.execute_input":"2024-02-06T04:42:59.035736Z","iopub.status.idle":"2024-02-06T04:42:59.591485Z","shell.execute_reply.started":"2024-02-06T04:42:59.035696Z","shell.execute_reply":"2024-02-06T04:42:59.590563Z"},"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":"2024-02-06T04:42:59.592721Z","iopub.execute_input":"2024-02-06T04:42:59.593205Z","iopub.status.idle":"2024-02-06T04:43:09.072753Z","shell.execute_reply.started":"2024-02-06T04:42:59.593136Z","shell.execute_reply":"2024-02-06T04:43:09.07144Z"},"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":"2024-02-06T04:43:09.074244Z","iopub.execute_input":"2024-02-06T04:43:09.074627Z","iopub.status.idle":"2024-02-06T04:43:09.08143Z","shell.execute_reply.started":"2024-02-06T04:43:09.074594Z","shell.execute_reply":"2024-02-06T04:43:09.079972Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"useless_features = [\"inittransactionamount_650A\", \"avglnamtstart24m_4525187A\",\"lastotherinc_902A\", \"lastotherlnsexpense_631A\",\"maxannuity_4075009A\",\"pmtaverage_3A\",\"pmtaverage_4527227A\", \"pmtaverage_4955615A\", \"totinstallast1m_4525188A\",\"maxpmtlast3m_4525190A\",\"avgpmtlast12m_4525200A\"]\nX_train = X_train.drop(useless_features,axis=1)\nX_valid = X_valid.drop(useless_features,axis=1)","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:09.083548Z","iopub.execute_input":"2024-02-06T04:43:09.084398Z","iopub.status.idle":"2024-02-06T04:43:09.292645Z","shell.execute_reply.started":"2024-02-06T04:43:09.084349Z","shell.execute_reply":"2024-02-06T04:43:09.291373Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train.isnull().sum()","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:09.294135Z","iopub.execute_input":"2024-02-06T04:43:09.294543Z","iopub.status.idle":"2024-02-06T04:43:09.366839Z","shell.execute_reply.started":"2024-02-06T04:43:09.29451Z","shell.execute_reply":"2024-02-06T04:43:09.365661Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"float64_cols = X_train.select_dtypes(include='float64').columns\nX_train[float64_cols] = X_train[float64_cols].astype('float32')\nX_valid[float64_cols] = X_valid[float64_cols].astype('float32')","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:09.371016Z","iopub.execute_input":"2024-02-06T04:43:09.371472Z","iopub.status.idle":"2024-02-06T04:43:09.697454Z","shell.execute_reply.started":"2024-02-06T04:43:09.371433Z","shell.execute_reply":"2024-02-06T04:43:09.696169Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train.dtypes","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:09.698564Z","iopub.execute_input":"2024-02-06T04:43:09.698915Z","iopub.status.idle":"2024-02-06T04:43:09.709633Z","shell.execute_reply.started":"2024-02-06T04:43:09.698885Z","shell.execute_reply":"2024-02-06T04:43:09.707906Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nfrom sklearn.impute import SimpleImputer\nfrom sklearn.preprocessing import OneHotEncoder\nfrom sklearn.model_selection import train_test_split\nimport numpy as np\n\n# Assuming X_train and X_valid are your dataframes with both numeric and categorical features\n\n# Numeric Data\nnumeric_imputer = SimpleImputer(strategy='mean')  # or 'median'\nX_train_numeric = numeric_imputer.fit_transform(X_train.select_dtypes(include=['float32', 'int32']))\nX_valid_numeric = numeric_imputer.transform(X_valid.select_dtypes(include=['float32', 'int32']))\n\n# Categorical Data\ncategorical_imputer = SimpleImputer(strategy='most_frequent')\nX_train_categorical = categorical_imputer.fit_transform(X_train.select_dtypes(include=['category']))\nX_valid_categorical = categorical_imputer.transform(X_valid.select_dtypes(include=['category']))\n\n# Step 2: Encoding Categorical Features\n\n# One-Hot Encoding\n#encoder = OneHotEncoder(handle_unknown='ignore', sparse=False)\n#X_train_encoded = encoder.fit_transform(X_train_categorical)\n#X_valid_encoded = encoder.transform(X_valid_categorical)\n\n# Creating DataFrames with Column Names\nnumeric_columns = X_train.select_dtypes(include=['float32', 'int32']).columns\n#categorical_columns = list(encoder.get_feature_names_out(X_train.select_dtypes(include=['category']).columns))\ncategorical_columns = X_train.select_dtypes(include=['category']).columns\n\nX_train_numeric_df = pd.DataFrame(X_train_numeric, columns=numeric_columns)\nX_valid_numeric_df = pd.DataFrame(X_valid_numeric, columns=numeric_columns)\n\nX_train_encoded_df = pd.DataFrame(X_train_categorical, columns=categorical_columns)\nX_valid_encoded_df = pd.DataFrame(X_valid_categorical, columns=categorical_columns)\n\n# Concatenating DataFrames\nX_train_final = pd.concat([X_train_numeric_df, X_train_encoded_df], axis=1)\nX_valid_final = pd.concat([X_valid_numeric_df, X_valid_encoded_df], axis=1)","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:09.7113Z","iopub.execute_input":"2024-02-06T04:43:09.712327Z","iopub.status.idle":"2024-02-06T04:43:14.632395Z","shell.execute_reply.started":"2024-02-06T04:43:09.71229Z","shell.execute_reply":"2024-02-06T04:43:14.631255Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train = X_train_final\nX_valid = X_valid_final","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:14.637759Z","iopub.execute_input":"2024-02-06T04:43:14.638806Z","iopub.status.idle":"2024-02-06T04:43:14.644467Z","shell.execute_reply.started":"2024-02-06T04:43:14.638746Z","shell.execute_reply":"2024-02-06T04:43:14.643204Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train.shape","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:14.646127Z","iopub.execute_input":"2024-02-06T04:43:14.646577Z","iopub.status.idle":"2024-02-06T04:43:14.660557Z","shell.execute_reply.started":"2024-02-06T04:43:14.64654Z","shell.execute_reply":"2024-02-06T04:43:14.659257Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from sklearn.decomposition import PCA\nfrom sklearn.preprocessing import StandardScaler, OneHotEncoder\nfrom sklearn.compose import ColumnTransformer\nfrom sklearn.pipeline import Pipeline\n\n# Assuming X_train and X_valid are your feature matrices\n# Assuming cat_cols is a list of the categorical column names\nn_components = 24\n\n# Separate numerical and categorical columns\nnum_cols = X_train.select_dtypes(include=['float32', 'int32']).columns\ncat_cols = X_train.select_dtypes(include=['category']).columns\n\n# Create a preprocessor to handle numerical and categorical features separately\npreprocessor = ColumnTransformer(\n    transformers=[\n        ('num', StandardScaler(), num_cols),\n        ('cat', OneHotEncoder(drop='first', sparse=False), cat_cols)\n    ]\n)\n\n# Create a pipeline with the preprocessor and PCA\npipeline = Pipeline([\n    ('preprocessor', preprocessor),\n    ('pca', PCA(n_components=n_components))\n])\n\n# Fit and transform the data\nX_train_pca = pipeline.fit_transform(X_train)\nX_valid_pca = pipeline.transform(X_valid)\n\n# Print the explained variance ratio for each principal component\nprint(\"Explained Variance Ratio:\")\nprint(pipeline.named_steps['pca'].explained_variance_ratio_)","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:14.661973Z","iopub.execute_input":"2024-02-06T04:43:14.662332Z","iopub.status.idle":"2024-02-06T04:43:15.96424Z","shell.execute_reply.started":"2024-02-06T04:43:14.662301Z","shell.execute_reply":"2024-02-06T04:43:15.962409Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_train = pd.DataFrame(X_train_pca) \nX_valid = pd.DataFrame(X_valid_pca) ","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:15.972567Z","iopub.execute_input":"2024-02-06T04:43:15.97427Z","iopub.status.idle":"2024-02-06T04:43:15.982877Z","shell.execute_reply.started":"2024-02-06T04:43:15.974175Z","shell.execute_reply":"2024-02-06T04:43:15.981025Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import matplotlib.pyplot as plt\nexplained_variance_ratio = pipeline.named_steps['pca'].explained_variance_ratio_\ncumulative_explained_variance = explained_variance_ratio.cumsum()\n\n# Plot cumulative explained variance\nplt.figure(figsize=(8, 6))\nplt.plot(range(1, n_components + 1), cumulative_explained_variance, marker='o', linestyle='-', color='b')\nplt.title('Cumulative Explained Variance by Principal Component')\nplt.xlabel('Number of Principal Components')\nplt.ylabel('Cumulative Explained Variance')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:43:15.986054Z","iopub.execute_input":"2024-02-06T04:43:15.988081Z","iopub.status.idle":"2024-02-06T04:43:16.359486Z","shell.execute_reply.started":"2024-02-06T04:43:15.988009Z","shell.execute_reply":"2024-02-06T04:43:16.358255Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training XGB\n\nMinimal example of XGB Classfier training is shown below.","metadata":{}},{"cell_type":"code","source":"import xgboost as xgb\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.metrics import roc_auc_score\n\n# Assuming you have your features X_train and labels y_train\n\n# Define the XGBoost model with your parameters\nmodel = xgb.XGBClassifier(\n    device=\"cuda\",\n    objective='binary:logistic',\n    tree_method=\"hist\",\n    enable_categorical=True,\n    eval_metric='auc',\n    subsample=1,\n    colsample_bytree=1,\n    min_child_weight=1,\n    max_depth=40,\n    n_estimators=1000,\n    random_state=42,\n)\n\n# Specify the number of folds for cross-validation\nn_splits = 10  # You can adjust the number of folds as needed\n\n# Initialize StratifiedKFold\nkf = StratifiedKFold(n_splits=n_splits, shuffle=True, random_state=42)\n\n# Lists to store the performance metrics for each fold\ntrain_auc_scores = []\nvalid_auc_scores = []\n\n# Perform k-fold cross-validation\nfor train_index, valid_index in kf.split(X_train, y_train):\n    X_train_fold, X_valid_fold = X_train.iloc[train_index], X_train.iloc[valid_index]\n    y_train_fold, y_valid_fold = y_train.iloc[train_index], y_train.iloc[valid_index]\n\n    # Train the model on the training fold\n    model.fit(\n        X_train_fold, y_train_fold,\n        eval_set=[(X_valid_fold, y_valid_fold)],\n        verbose=True, early_stopping_rounds=3\n    )\n\n    # Predict on the validation fold and calculate AUC\n    y_train_pred = model.predict_proba(X_train_fold)[:, 1]\n    train_auc = roc_auc_score(y_train_fold, y_train_pred)\n\n    y_valid_pred = model.predict_proba(X_valid_fold)[:, 1]\n    valid_auc = roc_auc_score(y_valid_fold, y_valid_pred)\n\n    # Store the AUC scores for later analysis\n    train_auc_scores.append(train_auc)\n    valid_auc_scores.append(valid_auc)\n\n# Print the average AUC scores across all folds\nprint(f\"Average Train AUC: {sum(train_auc_scores) / n_splits:.4f}\")\nprint(f\"Average Valid AUC: {sum(valid_auc_scores) / n_splits:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:44:20.478253Z","iopub.execute_input":"2024-02-06T04:44:20.478914Z","iopub.status.idle":"2024-02-06T04:48:47.908221Z","shell.execute_reply.started":"2024-02-06T04:44:20.478869Z","shell.execute_reply":"2024-02-06T04:48:47.906949Z"},"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":"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":"2024-02-06T04:48:59.897597Z","iopub.execute_input":"2024-02-06T04:48:59.898131Z","iopub.status.idle":"2024-02-06T04:49:01.298621Z","shell.execute_reply.started":"2024-02-06T04:48:59.898093Z","shell.execute_reply":"2024-02-06T04:49:01.297037Z"},"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)\n#categorical_cols = X_train.select_dtypes(include=['category']).columns\n\nfor col in cat_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 = model.predict(X_submission, num_iteration=gbm.best_iteration)","metadata":{"execution":{"iopub.status.busy":"2024-02-06T04:44:18.546486Z","iopub.status.idle":"2024-02-06T04:44:18.547549Z","shell.execute_reply.started":"2024-02-06T04:44:18.547312Z","shell.execute_reply":"2024-02-06T04:44:18.547342Z"},"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":"2024-02-06T04:44:18.549423Z","iopub.status.idle":"2024-02-06T04:44:18.549887Z","shell.execute_reply.started":"2024-02-06T04:44:18.549672Z","shell.execute_reply":"2024-02-06T04:44:18.549693Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}