ins-pricing 0.1.11__py3-none-any.whl → 0.2.0__py3-none-any.whl

ins_pricing/modelling/{bayesopt → core/bayesopt}/config_preprocess.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 import json
 import os
-from dataclasses import dataclass
+from dataclasses import dataclass, asdict
 from datetime import datetime
 from pathlib import Path
 from typing import Any, Dict, List, Optional
@@ -117,6 +117,36 @@ class BayesOptConfig:
     final_ensemble: bool = False
     final_ensemble_k: int = 3
     final_refit: bool = True
+    cv_strategy: str = "random"
+    cv_splits: Optional[int] = None
+    cv_group_col: Optional[str] = None
+    cv_time_col: Optional[str] = None
+    cv_time_ascending: bool = True
+    ft_oof_folds: Optional[int] = None
+    ft_oof_strategy: Optional[str] = None
+    ft_oof_shuffle: bool = True
+    save_preprocess: bool = False
+    preprocess_artifact_path: Optional[str] = None
+    plot_path_style: str = "nested"
+    bo_sample_limit: Optional[int] = None
+    cache_predictions: bool = False
+    prediction_cache_dir: Optional[str] = None
+    prediction_cache_format: str = "parquet"
+
+
+@dataclass
+class PreprocessArtifacts:
+    factor_nmes: List[str]
+    cate_list: List[str]
+    num_features: List[str]
+    var_nmes: List[str]
+    cat_categories: Dict[str, List[Any]]
+    dummy_columns: List[str]
+    numeric_scalers: Dict[str, Dict[str, float]]
+    weight_nme: str
+    resp_nme: str
+    binary_resp_nme: Optional[str] = None
+    drop_first: bool = True
 
 
 class OutputManager:
@@ -191,6 +221,7 @@ class DatasetPreprocessor:
         self.test_oht_scl_data: Optional[pd.DataFrame] = None
         self.var_nmes: List[str] = []
         self.cat_categories_for_shap: Dict[str, List[Any]] = {}
+        self.numeric_scalers: Dict[str, Dict[str, float]] = {}
 
     def run(self) -> "DatasetPreprocessor":
         """Run preprocessing: categorical encoding, target clipping, numeric scaling."""
@@ -291,6 +322,13 @@ class DatasetPreprocessor:
                 train_oht_scaled[num_chr].values.reshape(-1, 1))
             test_oht_scaled[num_chr] = scaler.transform(
                 test_oht_scaled[num_chr].values.reshape(-1, 1))
+            scale_val = float(getattr(scaler, "scale_", [1.0])[0])
+            if scale_val == 0.0:
+                scale_val = 1.0
+            self.numeric_scalers[num_chr] = {
+                "mean": float(getattr(scaler, "mean_", [0.0])[0]),
+                "scale": scale_val,
+            }
         # Fill missing dummy columns when reindexing to align train/test columns.
         test_oht_scaled = test_oht_scaled.reindex(
             columns=train_oht_scaled.columns, fill_value=0)
@@ -301,3 +339,28 @@ class DatasetPreprocessor:
             col for col in train_oht_scaled.columns if col not in excluded
         ]
         return self
+
+    def export_artifacts(self) -> PreprocessArtifacts:
+        dummy_columns: List[str] = []
+        if self.train_oht_data is not None:
+            dummy_columns = list(self.train_oht_data.columns)
+        return PreprocessArtifacts(
+            factor_nmes=list(self.config.factor_nmes),
+            cate_list=list(self.config.cate_list or []),
+            num_features=list(self.num_features),
+            var_nmes=list(self.var_nmes),
+            cat_categories=dict(self.cat_categories_for_shap),
+            dummy_columns=dummy_columns,
+            numeric_scalers=dict(self.numeric_scalers),
+            weight_nme=str(self.config.weight_nme),
+            resp_nme=str(self.config.resp_nme),
+            binary_resp_nme=self.config.binary_resp_nme,
+            drop_first=True,
+        )
+
+    def save_artifacts(self, path: str | Path) -> str:
+        payload = self.export_artifacts()
+        target = Path(path)
+        target.parent.mkdir(parents=True, exist_ok=True)
+        target.write_text(json.dumps(asdict(payload), ensure_ascii=True, indent=2), encoding="utf-8")
+        return str(target)