dragon-ml-toolbox 10.9.0__py3-none-any.whl → 10.10.1__py3-none-any.whl

{dragon_ml_toolbox-10.9.0.dist-info → dragon_ml_toolbox-10.10.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dragon-ml-toolbox
-Version: 10.9.0
+Version: 10.10.1
 Summary: A collection of tools for data science and machine learning projects.
 Author-email: Karl Loza <luigiloza@gmail.com>
 License-Expression: MIT

{dragon_ml_toolbox-10.9.0.dist-info → dragon_ml_toolbox-10.10.1.dist-info}/RECORD

@@ -1,21 +1,21 @@
-dragon_ml_toolbox-10.9.0.dist-info/licenses/LICENSE,sha256=2uUFNy7D0TLgHim1K5s3DIJ4q_KvxEXVilnU20cWliY,1066
-dragon_ml_toolbox-10.9.0.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=lY4_rJPnLnMu7YBQaY-_iz1JRDcLdQzNCyeLAF1glJY,1837
+dragon_ml_toolbox-10.10.1.dist-info/licenses/LICENSE,sha256=2uUFNy7D0TLgHim1K5s3DIJ4q_KvxEXVilnU20cWliY,1066
+dragon_ml_toolbox-10.10.1.dist-info/licenses/LICENSE-THIRD-PARTY.md,sha256=lY4_rJPnLnMu7YBQaY-_iz1JRDcLdQzNCyeLAF1glJY,1837
 ml_tools/ETL_cleaning.py,sha256=lSP5q6-ukGhJBPV8dlsqJvPXAzj4du_0J-SbtEd0Pjg,19292
 ml_tools/ETL_engineering.py,sha256=a6KCWH6kRatZtjaFEF_o917ApPMK5_vRD-BjfCDAl-E,49400
 ml_tools/GUI_tools.py,sha256=kEQWg-bog3pB5tI22gMGKWaCGHnz9TB2Lvvfhf5F2CI,45412
 ml_tools/MICE_imputation.py,sha256=kVSythWfxJFR4-2mtcYCWQaQ1Oz5yyx_SJu5gjnS7H8,11670
 ml_tools/ML_callbacks.py,sha256=JPvEw_cW5tYNJ2rMSgnNrKLuni_UrmuhDFaOw-u2SvA,13926
-ml_tools/ML_datasetmaster.py,sha256=BMmdCVAZ-HSnnSPLzKla2TdZKvHkHj4t9A0V1Ba3i-I,30821
-ml_tools/ML_evaluation.py,sha256=q4_RsBjmidc_yDX-DQvpJW8RCHrOCJbgXKBORQdt-TM,16111
+ml_tools/ML_datasetmaster.py,sha256=vqKZhCXsvN5yeRJdOKqMPh5OhY1xe6xlNjM3WoH5lys,30821
+ml_tools/ML_evaluation.py,sha256=6FB6S-aDDpFzQdrp3flBVECzEsHhMbQknYVGhHooEFs,16207
 ml_tools/ML_evaluation_multi.py,sha256=2jTSNFCu8cz5C05EusnrDyffs59M2Fq3UXSHxo2TR1A,12515
 ml_tools/ML_inference.py,sha256=SGDPiPxs_OYDKKRZziFMyaWcC8A37c70W9t-dMP5niI,23066
-ml_tools/ML_models.py,sha256=FliuqGhxP7AWHCweTLlfssXFOjwvFhIYJsgj_w_-EI4,27901
+ml_tools/ML_models.py,sha256=8UOMg9Qn8qtecUGfgnLRedX-lCWYwEs-C5RJ2m8mZM4,27544
 ml_tools/ML_optimization.py,sha256=a2Uxe1g-y4I-gFa8ENIM8QDS-Pz3hoPRRaVXAWMbyQA,13491
-ml_tools/ML_scaler.py,sha256=IrZsAr1xjvuLi8s5IKR-qbk2mS_awl3mn_xoXg5TJyA,7535
-ml_tools/ML_trainer.py,sha256=xw1zMgYpdqwsTt604xe3GTQNvpg6z6Ze-avmitGBFeU,23539
+ml_tools/ML_scaler.py,sha256=h2ymq5u953Lx60Qb38Y0mAWj85x9PbnP0xYNQ3pd8-w,7535
+ml_tools/ML_trainer.py,sha256=_g48w5Ak-wQr5fGHdJqlcpnzv3gWyL1ghkOhy9VOZbo,23930
 ml_tools/PSO_optimization.py,sha256=q0VYpssQGbPum7xdnkDXlJQKhZMYZo8acHpKhajPK3c,22954
 ml_tools/RNN_forecast.py,sha256=8rNZr-eWOBXMiDQV22e_tQTPM5LM2IFggEAa1FaoXaI,1965
-ml_tools/SQL.py,sha256=givoz6CGWRUdqnBem3VGZxzGdo3ZbX00kyHNjzI8kWE,10803
+ml_tools/SQL.py,sha256=rPeKywvwJ5oHYVUQUovO3OUkXQTxBT9Dvwb6E2ntphY,11233
 ml_tools/VIF_factor.py,sha256=MkMh_RIdsN2XUPzKNGRiEcmB17R_MmvGV4ezpL5zD2E,10403
 ml_tools/__init__.py,sha256=q0y9faQ6e17XCQ7eUiCZ1FJ4Bg5EQqLjZ9f_l5REUUY,41
 ml_tools/_logger.py,sha256=wcImAiXEZKPNcwM30qBh3t7HvoPURonJY0nrgMGF0sM,4719
@@ -30,7 +30,7 @@ ml_tools/keys.py,sha256=FDpbS3Jb0pjrVvvp2_8nZi919mbob_-xwuy5OOtKM_A,1848
 ml_tools/optimization_tools.py,sha256=P3I6lIpvZ8Xf2kX5FvvBKBmrK2pB6idBpkTzfUJxTeE,5073
 ml_tools/path_manager.py,sha256=wLJlz3Y9_1-LB9em4B2VYDCVuTOX2eOc7D6hbbebjgM,14990
 ml_tools/utilities.py,sha256=30z0x1aDLyBGzF98_tgSaxwFafYwQS-GTFzXHopBSGc,29105
-dragon_ml_toolbox-10.9.0.dist-info/METADATA,sha256=NK8z4StYOVR0ByF_l-vNjyrFgbb2qddBa6lOzlQsZrg,6968
-dragon_ml_toolbox-10.9.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-dragon_ml_toolbox-10.9.0.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
-dragon_ml_toolbox-10.9.0.dist-info/RECORD,,
+dragon_ml_toolbox-10.10.1.dist-info/METADATA,sha256=uwfI1-UU2czHm1VT4SoSZNrZur0DL2PWI23VUx9HaBM,6969
+dragon_ml_toolbox-10.10.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+dragon_ml_toolbox-10.10.1.dist-info/top_level.txt,sha256=wm-oxax3ciyez6VoO4zsFd-gSok2VipYXnbg3TH9PtU,9
+dragon_ml_toolbox-10.10.1.dist-info/RECORD,,

ml_tools/ML_datasetmaster.py

@@ -200,7 +200,7 @@ class _BaseDatasetMaker(ABC):
         filepath = save_path / filename
         self.scaler.save(filepath, verbose=False)
         if verbose:
-            _LOGGER.info(f"Scaler for dataset '{self.id}' saved to '{filepath.name}'.")
+            _LOGGER.info(f"Scaler for dataset '{self.id}' saved as '{filepath.name}'.")
 
 
 # Single target dataset

ml_tools/ML_evaluation.py

@@ -353,7 +353,7 @@ def shap_summary_plot(model,
     plt.ion()
 
 
-def plot_attention_importance(weights: List[torch.Tensor], feature_names: Optional[List[str]], save_dir: Union[str, Path]):
+def plot_attention_importance(weights: List[torch.Tensor], feature_names: Optional[List[str]], save_dir: Union[str, Path], top_n: int = 10):
     """
     Aggregates attention weights and plots global feature importance.
 
@@ -364,6 +364,7 @@ def plot_attention_importance(weights: List[torch.Tensor], feature_names: Option
         weights (List[torch.Tensor]): A list of attention weight tensors from each batch.
         feature_names (List[str] | None): Names of the features for plot labeling.
         save_dir (str | Path): Directory to save the plot and summary CSV.
+        top_n (int): The number of top features to display in the plot.
     """
     if not weights:
         _LOGGER.error("Attention weights list is empty. Skipping importance plot.")
@@ -392,11 +393,10 @@ def plot_attention_importance(weights: List[torch.Tensor], feature_names: Option
     summary_df.to_csv(summary_path, index=False)
     _LOGGER.info(f"📝 Attention summary data saved as '{summary_path.name}'")
 
-    # --- Step 3: Create and save the plot ---
-    plt.figure(figsize=(10, 8), dpi=100)
+    # --- Step 3: Create and save the plot for top N features ---
+    plot_df = summary_df.head(top_n).sort_values('mean_attention', ascending=True)
     
-    # Sort for plotting
-    plot_df = summary_df.sort_values('mean_attention', ascending=True)
+    plt.figure(figsize=(10, 8), dpi=100)
 
     # Create horizontal bar plot with error bars
     plt.barh(
@@ -410,7 +410,7 @@ def plot_attention_importance(weights: List[torch.Tensor], feature_names: Option
         color='cornflowerblue'
     )
     
-    plt.title('Global Feature Importance')
+    plt.title('Top Features by Attention')
     plt.xlabel('Average Attention Weight')
     plt.ylabel('Feature')
     plt.grid(axis='x', linestyle='--', alpha=0.6)

ml_tools/ML_models.py

@@ -43,7 +43,7 @@ class _ArchitectureHandlerMixin:
             json.dump(config, f, indent=4)
         
         if verbose:
-            _LOGGER.info(f"Architecture for '{self.__class__.__name__}' saved to '{path_dir.name}'")
+            _LOGGER.info(f"Architecture for '{self.__class__.__name__}' saved as '{full_path.name}'")
 
     @classmethod
     def load(cls: type, file_or_dir: Union[str, Path], verbose: bool = True) -> nn.Module:
@@ -147,6 +147,30 @@ class _BaseMLP(nn.Module, _ArchitectureHandlerMixin):
         return f"{name}(arch: {arch_str})"
 
 
+class _BaseAttention(_BaseMLP):
+    """
+    Abstract base class for MLP models that incorporate an attention mechanism
+    before the main MLP layers.
+    """
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        # By default, models inheriting this do not have the flag.
+        self.has_interpretable_attention = False
+    
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        """Defines the standard forward pass."""
+        logits, _attention_weights = self.forward_attention(x)
+        return logits
+
+    def forward_attention(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        """Returns logits and attention weights."""
+        # This logic is now shared and defined in one place
+        x, attention_weights = self.attention(x)
+        x = self.mlp(x)
+        logits = self.output_layer(x)
+        return logits, attention_weights
+
+
 class MultilayerPerceptron(_BaseMLP):
     """
     Creates a versatile Multilayer Perceptron (MLP) for regression or classification tasks.
@@ -184,7 +208,7 @@ class MultilayerPerceptron(_BaseMLP):
         return self._repr_helper(name="MultilayerPerceptron", mlp_layers=layer_sizes)
 
 
-class AttentionMLP(_BaseMLP):
+class AttentionMLP(_BaseAttention):
     """
     A Multilayer Perceptron (MLP) that incorporates an Attention layer to dynamically weigh input features.
     
@@ -205,25 +229,7 @@ class AttentionMLP(_BaseMLP):
         super().__init__(in_features, out_targets, hidden_layers, drop_out)
         # Attention
         self.attention = _AttentionLayer(in_features)
-
-    def forward(self, x: torch.Tensor) -> torch.Tensor:
-        """
-        Defines the standard forward pass.
-        """
-        logits, _attention_weights = self.forward_attention(x)
-        return logits
-    
-    def forward_attention(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
-        """
-        Returns logits and attention weights
-        """
-        # The attention layer returns the processed x and the weights
-        x, attention_weights = self.attention(x)
-        
-        # Pass the attention-modified tensor through the MLP
-        logits = self.mlp(x)
-        
-        return logits, attention_weights
+        self.has_interpretable_attention = True
     
     def __repr__(self) -> str:
         """Returns the developer-friendly string representation of the model."""
@@ -238,7 +244,7 @@ class AttentionMLP(_BaseMLP):
         return self._repr_helper(name="AttentionMLP", mlp_layers=arch)
 
 
-class MultiHeadAttentionMLP(_BaseMLP):
+class MultiHeadAttentionMLP(_BaseAttention):
     """
     An MLP that incorporates a standard `nn.MultiheadAttention` layer to process
     the input features.
@@ -267,24 +273,6 @@ class MultiHeadAttentionMLP(_BaseMLP):
             dropout=attention_dropout
         )
 
-    def forward(self, x: torch.Tensor) -> torch.Tensor:
-        """Defines the standard forward pass of the model."""
-        logits, _attention_weights = self.forward_attention(x)
-        return logits
-    
-    def forward_attention(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
-        """
-        Returns logits and attention weights.
-        """
-        # The attention layer returns the processed x and the weights
-        x, attention_weights = self.attention(x)
-        
-        # Pass the attention-modified tensor through the MLP and prediction head
-        x = self.mlp(x)
-        logits = self.output_layer(x)
-        
-        return logits, attention_weights
-
     def get_architecture_config(self) -> Dict[str, Any]:
         """Returns the full configuration of the model."""
         config = super().get_architecture_config()

ml_tools/ML_scaler.py

@@ -164,7 +164,7 @@ class PytorchScaler:
         }
         torch.save(state, path_obj)
         if verbose:
-            _LOGGER.info(f"PytorchScaler state saved to '{path_obj.name}'.")
+            _LOGGER.info(f"PytorchScaler state saved as '{path_obj.name}'.")
 
     @staticmethod
     def load(filepath: Union[str, Path], verbose: bool=True) -> 'PytorchScaler':

ml_tools/ML_trainer.py

@@ -472,23 +472,30 @@ class MLTrainer:
 
                 yield attention_weights
     
-    def explain_attention(self, save_dir: Union[str, Path], feature_names: Optional[List[str]], explain_dataset: Optional[Dataset] = None):
+    def explain_attention(self, save_dir: Union[str, Path], 
+                          feature_names: Optional[List[str]], 
+                          explain_dataset: Optional[Dataset] = None,
+                          plot_n_features: int = 10):
         """
         Generates and saves a feature importance plot based on attention weights.
 
-        This method only works for models with a `forward_attention` method.
+        This method only works for models with models with 'has_interpretable_attention'.
 
         Args:
             save_dir (str | Path): Directory to save the plot and summary data.
-            feature_names (List[str] | None): Names for the features for plot labeling.
+            feature_names (List[str] | None): Names for the features for plot labeling. If not given, generic names will be used.
             explain_dataset (Dataset, optional): A specific dataset to explain. If None, the trainer's test dataset is used.
+            plot_n_features (int): Number of top features to plot.
         """
         
         print("\n--- Attention Analysis ---")
         
         # --- Step 1: Check if the model supports this explanation ---
-        if not hasattr(self.model, 'forward_attention'):
-            _LOGGER.error("Model does not have a `forward_attention` method. Skipping attention explanation.")
+        if not getattr(self.model, 'has_interpretable_attention', False):
+            _LOGGER.warning(
+                "Model is not flagged for interpretable attention analysis. "
+                "Skipping. This is the correct behavior for models like MultiHeadAttentionMLP."
+            )
             return
 
         # --- Step 2: Set up the dataloader ---
@@ -514,7 +521,8 @@ class MLTrainer:
             plot_attention_importance(
                 weights=all_weights,
                 feature_names=feature_names,
-                save_dir=save_dir
+                save_dir=save_dir,
+                top_n=plot_n_features
             )
         else:
             _LOGGER.error("No attention weights were collected from the model.")

ml_tools/SQL.py

@@ -120,12 +120,14 @@ class DatabaseManager:
         if not self.cursor:
             _LOGGER.error("Database connection is not open.")
             raise sqlite3.Error()
+        
+        sanitized_table_name = sanitize_filename(table_name)
 
         columns = ', '.join(f'"{k}"' for k in data.keys())
         placeholders = ', '.join(['?'] * len(data))
         values = list(data.values())
         
-        query = f"INSERT INTO {table_name} ({columns}) VALUES ({placeholders})"
+        query = f'INSERT INTO "{sanitized_table_name}" ({columns}) VALUES ({placeholders})'
         
         self.cursor.execute(query, values)
 
@@ -187,6 +189,8 @@ class DatabaseManager:
         if not data:
             _LOGGER.warning("'insert_many' called with empty data list. No action taken.")
             return
+        
+        sanitized_table_name = sanitize_filename(table_name)
 
         # Assume all dicts have the same keys as the first one
         first_row = data[0]
@@ -196,10 +200,10 @@ class DatabaseManager:
         # Create a list of tuples, where each tuple is a row of values
         values_to_insert = [list(row.values()) for row in data]
 
-        query = f"INSERT INTO {table_name} ({columns}) VALUES ({placeholders})"
+        query = f'INSERT INTO "{sanitized_table_name}" ({columns}) VALUES ({placeholders})'
         
         self.cursor.executemany(query, values_to_insert)
-        _LOGGER.info(f"➡️ Bulk inserted {len(values_to_insert)} rows into '{table_name}'.")
+        _LOGGER.info(f"➡️ Bulk inserted {len(values_to_insert)} rows into '{sanitized_table_name}'.")
         
     def insert_from_dataframe(self, table_name: str, df: pd.DataFrame, if_exists: Literal['fail', 'replace', 'append'] = 'append'):
         """
@@ -220,9 +224,11 @@ class DatabaseManager:
         if not self.conn:
             _LOGGER.error("Database connection is not open.")
             raise sqlite3.Error()
+        
+        sanitized_table_name = sanitize_filename(table_name)
 
         df.to_sql(
-            table_name,
+            sanitized_table_name,
             self.conn,
             if_exists=if_exists,
             index=False  # Typically, we don't want to save the DataFrame index
@@ -248,9 +254,11 @@ class DatabaseManager:
         if not self.conn:
             _LOGGER.error("Database connection is not open.")
             raise sqlite3.Error()
+        
+        sanitized_table_name = sanitize_filename(table_name)
             
         # PRAGMA is a special SQL command in SQLite for database metadata
-        return pd.read_sql_query(f'PRAGMA table_info("{table_name}");', self.conn)
+        return pd.read_sql_query(f'PRAGMA table_info("{sanitized_table_name}");', self.conn)
 
     def create_index(self, table_name: str, column_name: str, unique: bool = False):
         """
@@ -269,11 +277,13 @@ class DatabaseManager:
         if not self.cursor:
             _LOGGER.error("Database connection is not open.")
             raise sqlite3.Error()
+        
+        sanitized_table_name = sanitize_filename(table_name)
 
-        index_name = f"idx_{table_name}_{column_name}"
+        index_name = f"idx_{sanitized_table_name}_{column_name}"
         unique_clause = "UNIQUE" if unique else ""
         
-        query = f"CREATE {unique_clause} INDEX IF NOT EXISTS {index_name} ON {table_name} ({column_name})"
+        query = f'CREATE {unique_clause} INDEX IF NOT EXISTS "{index_name}" ON "{sanitized_table_name}" ("{column_name}")'
         
         _LOGGER.info(f"➡️ Executing: {query}")
         self.cursor.execute(query)