dragon-ml-toolbox 20.12.0__py3-none-any.whl → 20.14.0__py3-none-any.whl

{dragon_ml_toolbox-20.12.0.dist-info → dragon_ml_toolbox-20.14.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dragon-ml-toolbox
-Version: 20.12.0
+Version: 20.14.0
 Summary: Complete pipelines and helper tools for data science and machine learning projects.
 Author-email: Karl Luigi Loza Vidaurre <luigiloza@gmail.com>
 License-Expression: MIT

{dragon_ml_toolbox-20.12.0.dist-info → dragon_ml_toolbox-20.14.0.dist-info}/RECORD

@@ -1,5 +1,5 @@
-dragon_ml_toolbox-20.12.0.dist-info/licenses/LICENSE,sha256=L35WDmmLZNTlJvxF6Vy7Uy4SYNi6rCfWUqlTHpoRMoU,1081
-dragon_ml_toolbox-20.12.0.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=0-HBRMMgKuwtGy6nMJZvIn1fLxhx_ksyyVB2U_iyYZU,2818
+dragon_ml_toolbox-20.14.0.dist-info/licenses/LICENSE,sha256=L35WDmmLZNTlJvxF6Vy7Uy4SYNi6rCfWUqlTHpoRMoU,1081
+dragon_ml_toolbox-20.14.0.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=0-HBRMMgKuwtGy6nMJZvIn1fLxhx_ksyyVB2U_iyYZU,2818
 ml_tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ml_tools/constants.py,sha256=3br5Rk9cL2IUo638eJuMOGdbGQaWssaUecYEvSeRBLM,3322
 ml_tools/ETL_cleaning/__init__.py,sha256=gLRHF-qzwpqKTvbbn9chIQELeUDh_XGpBRX28j-5IqI,545
@@ -46,7 +46,7 @@ ml_tools/ML_evaluation/_loss.py,sha256=1a4O25i3Ya_3naNZNL7ELLUL46BY86g1scA7d7q2U
 ml_tools/ML_evaluation/_regression.py,sha256=UZA7_fg85ZKJQWszioWDtmkplSiXeHJk2fBYR5bRXHY,11225
 ml_tools/ML_evaluation/_sequence.py,sha256=gUk9Uvmy7MrXkfrriMnfypkgJU5XERHdqekTa2gBaOM,8004
 ml_tools/ML_evaluation/_vision.py,sha256=abBHQ6Z2GunHNusL3wcLgfI1FVNA6hBUBTq1eOA8FSA,11489
-ml_tools/ML_evaluation_captum/_ML_evaluation_captum.py,sha256=eWxd6HRQSaqFYNek6x1hZBKi8qeHqk4oOkqrjQgKjZk,19611
+ml_tools/ML_evaluation_captum/_ML_evaluation_captum.py,sha256=eCP19o4sxfG0XlAVKiuuIxdtxO5lqCc0SuhWXx6eObY,20079
 ml_tools/ML_evaluation_captum/__init__.py,sha256=DZDoZXexCI49JNl_tTmFfYW4hTUYK5QQLex01wMfhnk,333
 ml_tools/ML_finalize_handler/_ML_finalize_handler.py,sha256=g-vkHJDTGXZsKOUA-Yfg7EuA1SmaHjzesCPiAyRMg2k,7054
 ml_tools/ML_finalize_handler/__init__.py,sha256=VQyLbCQUcliAAFiOAsnPhyJ7UVYgbSqAbAnpqeOnRSg,198
@@ -80,7 +80,7 @@ ml_tools/ML_optimization/__init__.py,sha256=No18Dsw6Q9zPt8B9fpG0bWomuXmwDC7Dioki
 ml_tools/ML_optimization/_multi_dragon.py,sha256=zQhDxFY8FNxUlcbSnHMVArfojzYjgNa21jSE3pJmRW0,38956
 ml_tools/ML_optimization/_single_dragon.py,sha256=jh5-SK6NKAzbheQhquiYoROozk-RzUv1jiFkIzK_AFg,7288
 ml_tools/ML_optimization/_single_manual.py,sha256=h-_k9JmRqPkjTra1nu7AyYbSyWkYZ1R3utiNmW06WFs,21809
-ml_tools/ML_scaler/_ML_scaler.py,sha256=P75X0Sx8N-VxC2Qy8aG7mWaZlkTfjspiZDi1YiMQD1I,8872
+ml_tools/ML_scaler/_ML_scaler.py,sha256=NcwprqrAHMIKpkzMdExk99I2QpfTSbiJH8rDqmOlnkU,8870
 ml_tools/ML_scaler/__init__.py,sha256=SHDNyLsoOLl2OtkIb3pGg-JRs3E2bYJBgnHwH3vw_Tk,172
 ml_tools/ML_trainer/__init__.py,sha256=42kueHa7Z0b_yLbywNCgIxlW6WmgLBqkTFwKH7vFLXw,379
 ml_tools/ML_trainer/_base_trainer.py,sha256=0ATm672NRsjJ6nv_NEl6-OEd9Bst1-s5OPxfG4qe8Lg,18075
@@ -104,10 +104,10 @@ ml_tools/_core/__init__.py,sha256=m-VP0RW0tOTm9N5NI3kFNcpM7WtVgs0RK9pK3ZJRZQQ,14
 ml_tools/_core/_logger.py,sha256=xzhn_FouMDRVNwXGBGlPC9Ruq6i5uCrmNaS5jesguMU,4972
 ml_tools/_core/_schema_load_ops.py,sha256=KLs9vBzANz5ESe2wlP-C41N4VlgGil-ywcfvWKSOGss,1551
 ml_tools/_core/_script_info.py,sha256=LtFGt10gEvCnhIRMKJPi2yXkiGLcdr7lE-oIP2XGHzQ,234
-ml_tools/data_exploration/__init__.py,sha256=efUBsruHL56B429tUadl3PdG73zAF639Y430uMQRfko,1917
-ml_tools/data_exploration/_analysis.py,sha256=PJNrEBz5ZZXHoUlQ6fh9Y86nzPQrLpVPv2Ye4NfOxgs,14181
+ml_tools/data_exploration/__init__.py,sha256=XNA8gcRx5ifrv092HA7HSpek8havlk_3RZi9aq9dSjg,1957
+ml_tools/data_exploration/_analysis.py,sha256=JSoFJSkv4-_v9YxxmjHZ_PeFRneDENjSEo2sy_uC4oY,14196
 ml_tools/data_exploration/_cleaning.py,sha256=pAZOXgGK35j7O8q6cnyTwYK1GLNnD04A8p2fSyMB1mg,20906
-ml_tools/data_exploration/_features.py,sha256=Z1noJfDxBzFRfusFp6NlpLF2NItuZuzFHq4ssWFqny4,26273
+ml_tools/data_exploration/_features.py,sha256=twJ6OixU4ItRXA8rPJRfg2N9QVsbn38CFqJiLcXav1A,28664
 ml_tools/data_exploration/_plotting.py,sha256=zH1dPcIoAlOuww23xIoBCsQOAshPPv9OyGposOA2RvI,19883
 ml_tools/data_exploration/_schema_ops.py,sha256=Fd6fBGGv4OpxmJ1HG9pith6QL90z0tzssCvzkQxlEEQ,11083
 ml_tools/ensemble_evaluation/__init__.py,sha256=t4Gr8EGEk8RLatyc92-S0BzbQvdvodzoF-qDAH2qjVg,546
@@ -119,7 +119,7 @@ ml_tools/ensemble_learning/_ensemble_learning.py,sha256=MHDZBR20_nStlSSeThFI3bSu
 ml_tools/excel_handler/__init__.py,sha256=AaWM3n_dqBhJLTs3OEA57ex5YykKXNOwVCyHlVsdnqI,530
 ml_tools/excel_handler/_excel_handler.py,sha256=TODudmeQgDSdxUKzLfAzizs--VL-g8WxDOfQ4sgxxLs,13965
 ml_tools/keys/__init__.py,sha256=-0c2pmrhyfROc-oQpEjJGLBMhSagA3CyFijQaaqZRqU,399
-ml_tools/keys/_keys.py,sha256=Kr73o9SaH5Y3DT0z0H-1eLwlBplJmjisjoO_EoUNkAg,9388
+ml_tools/keys/_keys.py,sha256=YE_Ux2FYObfWurcQvfCvA3ZehwOvKvtCvIViUuYAYNM,9447
 ml_tools/math_utilities/__init__.py,sha256=K7Obkkc4rPKj4EbRZf1BsXHfiCg7FXYv_aN9Yc2Z_Vg,400
 ml_tools/math_utilities/_math_utilities.py,sha256=BYHIVcM9tuKIhVrkgLLiM5QalJ39zx7dXYy_M9aGgiM,9012
 ml_tools/optimization_tools/__init__.py,sha256=KD8JXpfGuPndO4AHnjJGu6uV1GRwhOfboD0KZV45kzw,658
@@ -143,7 +143,7 @@ ml_tools/utilities/__init__.py,sha256=h4lE3SQstg-opcQj6QSKhu-HkqSbmHExsWoM9vC5D9
 ml_tools/utilities/_translate.py,sha256=U8hRPa3PmTpIf9n9yR3gBGmp_hkcsjQLwjAHSHc0WHs,10325
 ml_tools/utilities/_utility_save_load.py,sha256=EFvFaTaHahDQWdJWZr-j7cHqRbG_Xrpc96228JhV-bs,16773
 ml_tools/utilities/_utility_tools.py,sha256=bN0J9d1S0W5wNzNntBWqDsJcEAK7-1OgQg3X2fwXns0,6918
-dragon_ml_toolbox-20.12.0.dist-info/METADATA,sha256=VH-wt974dX5kNOfVShO_N5HJJCIhqP2V_7uihSwlYzE,7889
-dragon_ml_toolbox-20.12.0.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-dragon_ml_toolbox-20.12.0.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
-dragon_ml_toolbox-20.12.0.dist-info/RECORD,,
+dragon_ml_toolbox-20.14.0.dist-info/METADATA,sha256=32IleSQa7t7E42ZB5rM32Lf1MlSAMtKkU-TFky3VckA,7889
+dragon_ml_toolbox-20.14.0.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+dragon_ml_toolbox-20.14.0.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
+dragon_ml_toolbox-20.14.0.dist-info/RECORD,,

ml_tools/ML_evaluation_captum/_ML_evaluation_captum.py

@@ -202,6 +202,13 @@ def _process_single_target(ig: 'IntegratedGradients', # type: ignore
         mean_abs_attr = mean_abs_attr[:min_len]
         feature_names = feature_names[:min_len]
     
+    # Calculate percentages (Before Min-Max scaling to preserve relative importance)
+    total_attr_sum = np.sum(mean_abs_attr)
+    if total_attr_sum > 0:
+        attr_percentages = (mean_abs_attr / total_attr_sum) * 100.0
+    else:
+        attr_percentages = np.zeros_like(mean_abs_attr)
+    
     # Min-Max Scaling
     target_min = 0.01
     target_max = 1.0
@@ -222,7 +229,8 @@ def _process_single_target(ig: 'IntegratedGradients', # type: ignore
     # --- Save Data to CSV ---
     summary_df = pd.DataFrame({
         CaptumKeys.FEATURE_COLUMN: feature_names,
-        CaptumKeys.IMPORTANCE_COLUMN: mean_abs_attr
+        CaptumKeys.IMPORTANCE_COLUMN: mean_abs_attr,
+        CaptumKeys.PERCENT_COLUMN: attr_percentages
     }).sort_values(CaptumKeys.IMPORTANCE_COLUMN, ascending=False)
     
     csv_name = f"{CaptumKeys.SAVENAME}{file_suffix}.csv"
@@ -230,11 +238,13 @@ def _process_single_target(ig: 'IntegratedGradients', # type: ignore
     summary_df.to_csv(csv_path, index=False)
 
     # --- Generate Plot ---
-    plot_df = summary_df.head(20).sort_values(CaptumKeys.IMPORTANCE_COLUMN, ascending=True)
+    plot_df = summary_df.head(20).sort_values(CaptumKeys.PERCENT_COLUMN, ascending=True)
     plt.figure(figsize=(10, 8), dpi=300)
-    plt.barh(plot_df[CaptumKeys.FEATURE_COLUMN], plot_df[CaptumKeys.IMPORTANCE_COLUMN], color='mediumpurple')
-    plt.xlim(0, 1.05) # standardized scale
-    plt.xlabel("Mean Absolute Attribution")
+    plt.barh(plot_df[CaptumKeys.FEATURE_COLUMN], plot_df[CaptumKeys.PERCENT_COLUMN], color='mediumpurple')
+    # plt.xlim(0, 1.05) # standardized scale # Removed to reflect actual percentages
+    plt.xlim(left=0) # start at 0
+    # plt.xlabel("Scaled Mean Absolute Attribution")
+    plt.xlabel("Relative Importance (%)")
     
     title = "Feature Importance"
     

ml_tools/ML_scaler/_ML_scaler.py

@@ -99,7 +99,7 @@ class DragonScaler:
             std = torch.sqrt(torch.clamp(variance, min=1e-8))
         
         if verbose >= 2:
-            _LOGGER.info(f"Scaler fitted on {n_total} samples for {num_continuous_features} features (Welford's).")
+            _LOGGER.info(f"Scaler fitted on {n_total} samples for {num_continuous_features} columns (Welford's).")
         return cls(mean=mean_global, std=std, continuous_feature_indices=continuous_feature_indices)
 
     @classmethod
@@ -121,7 +121,7 @@ class DragonScaler:
         std = torch.where(std == 0, torch.tensor(1.0, device=data.device), std)
         
         if verbose >= 2:
-            _LOGGER.info(f"Scaler fitted on tensor with {data.shape[0]} samples for {num_features} features.")
+            _LOGGER.info(f"Scaler fitted on tensor with {data.shape[0]} samples for {num_features} columns.")
         
         return cls(mean=mean, std=std, continuous_feature_indices=indices)
 

ml_tools/data_exploration/__init__.py

@@ -33,6 +33,7 @@ from ._features import (
     reconstruct_one_hot,
     reconstruct_binary,
     reconstruct_multibinary,
+    filter_subset,
 )
 
 from ._schema_ops import (
@@ -51,6 +52,7 @@ __all__ = [
     "drop_columns_with_missing_data",
     "drop_macro",
     "clean_column_names",
+    "filter_subset",
     "plot_value_distributions",
     "split_features_targets", 
     "split_continuous_binary", 

ml_tools/data_exploration/_analysis.py

@@ -34,7 +34,7 @@ def summarize_dataframe(df: pd.DataFrame, round_digits: int = 2):
     """
     summary = pd.DataFrame({
         'Data Type': df.dtypes,
-        'Completeness %': (df.notnull().mean() * 100).round(2),
+        'Completeness %': (df.notnull().mean() * 100).round(2), # type: ignore
         'Unique Values': df.nunique(),
         # 'Missing %': (df.isnull().mean() * 100).round(2)
     })

ml_tools/data_exploration/_features.py

@@ -657,3 +657,66 @@ def reconstruct_multibinary(
         _LOGGER.info(f"Reconstructed {converted_count} binary columns matching '{pattern}'.")
 
     return new_df, target_columns
+
+
+def filter_subset(
+    df: pd.DataFrame,
+    filters: Union[dict[str, Any], dict[str, list[Any]]],
+    drop_filter_cols: bool = True,
+    reset_index: bool = True,
+    verbose: int = 3
+) -> pd.DataFrame:
+    """
+    Filters the DataFrame based on a dictionary of column-value conditions.
+
+    Supports:
+    - Single value matching (e.g., {"Color": "Blue"})
+    - Multiple value matching (e.g., {"Color": ["Blue", "Red"]}) -> OR logic within column.
+    - Multiple column filtering (e.g., {"Color": "Blue", "Size": "Large"}) -> AND logic between columns.
+
+    Args:
+        df (pd.DataFrame): Input DataFrame.
+        filters (dict[str, Any] | dict[str, list[Any]]): Dictionary where keys are column names and values are the target values (scalar or list).
+        drop_filter_cols (bool): If True, drops the columns used for filtering from the result.
+        reset_index (bool): If True, resets the index of the resulting DataFrame.
+        verbose (int): Verbosity level.
+
+    Returns:
+        pd.DataFrame: The filtered DataFrame.
+    """
+    df_filtered = df.copy()
+
+    # Validate columns exist
+    missing_cols = [col for col in filters.keys() if col not in df.columns]
+    if missing_cols:
+        _LOGGER.error(f"Filter columns not found: {missing_cols}")
+        raise ValueError()
+
+    if verbose >= 2:
+        _LOGGER.info(f"Original shape: {df.shape}")
+
+    for col, value in filters.items():
+        # Handle list of values (OR logic within column)
+        if isinstance(value, list):
+            df_filtered = df_filtered[df_filtered[col].isin(value)]
+        # Handle single value
+        else:
+            # Warn if the value is a floating point due to potential precision issues
+            if isinstance(value, float) and verbose >= 1:
+                _LOGGER.warning(f"Filtering on column '{col}' with float value '{value}'.")
+            df_filtered = df_filtered[df_filtered[col] == value]
+
+    if drop_filter_cols:
+        if verbose >= 3:
+            _LOGGER.info(f"Dropping filter columns: {list(filters.keys())}")
+        df_filtered.drop(columns=list(filters.keys()), inplace=True)
+
+    if reset_index:
+        if verbose >= 3:
+            _LOGGER.info("Resetting index of the filtered DataFrame.")
+        df_filtered.reset_index(drop=True, inplace=True)
+
+    if verbose >= 2:
+        _LOGGER.info(f"Filtered shape: {df_filtered.shape}")
+
+    return df_filtered

ml_tools/keys/_keys.py

@@ -99,8 +99,9 @@ class SHAPKeys:
 
 class CaptumKeys:
     """Keys for Captum functions"""
-    FEATURE_COLUMN = "feature"
-    IMPORTANCE_COLUMN = "importance"
+    FEATURE_COLUMN = "Feature"
+    IMPORTANCE_COLUMN = "Scaled Mean Attribution"
+    PERCENT_COLUMN = "Relative Importance(%)"
     SAVENAME = "captum_summary"
     PLOT_NAME = "captum_importance_plot"