fix(Pipeline): remove PCA step that introduced clear lookahead bias (#164)

data_loader/process_data.py

@@ -10,18 +10,11 @@ import warnings
 
 
 
-def process_data(X:pd.DataFrame, y:pd.Series, configs: dict) -> tuple[pd.DataFrame,pd.DataFrame,pd.DataFrame]:
+def process_data(X:pd.DataFrame, y:pd.Series, configs: dict) -> tuple[pd.DataFrame,pd.DataFrame]:
     
     model_config, training_config, data_config = itemgetter('model_config', 'training_config', 'data_config')(configs)
     
     original_X = X.copy()
-
-    # 2a. Dimensionality Reduction (optional)
-    if training_config['dimensionality_reduction']:
-        X_pca = reduce_dimensionality(X, int(len(X.columns) / 2))
-        X = X_pca.copy()
-    else:
-        X_pca = X.copy()
     
     # 2b. Feature Selection
     print("Feature Selection started")
@@ -29,7 +22,7 @@ def process_data(X:pd.DataFrame, y:pd.Series, configs: dict) -> tuple[pd.DataFra
     backup_model = default_feature_selector_regression if data_config['method'] == 'regression' else default_feature_selector_classification
     X = select_features(X = X, y = y, model = model_config['primary_models'][0][1], n_features_to_select = training_config['n_features_to_select'], backup_model = backup_model, scaling = training_config['scaler'])
     
-    return X, original_X, X_pca
+    return X, original_X
 
 def check_data(X:pd.DataFrame, y:pd.Series, training_config:dict):
     """ Returns True if data is valid, else returns False."""

run_pipeline.py

@@ -56,13 +56,13 @@ def __run_training(model_config:dict, training_config:dict, data_config:dict):
         # 1. Load data, check for validity and process data (feature selection, dimensionality reduction, etc.)
         X, y, target_returns = load_data(**configs['data_config'])
         if check_data(X, y, training_config) is False: continue
-        X, original_X, X_pca = process_data(X, y, configs)
+        X, original_X = process_data(X, y, configs)
 
         # 2. Train a Primary model with optional metalabeling for each asset
-        training_step_primary, current_predictions = primary_step(X, y, original_X, X_pca, asset, target_returns, configs, reporting)
+        training_step_primary, current_predictions = primary_step(X, y, original_X, asset, target_returns, configs, reporting)
         
         # 3. Train an Ensemble model with optional metalabeling for each asset
-        training_step_secondary = secondary_step(X, y, original_X, X_pca, current_predictions, asset, target_returns, configs, reporting)
+        training_step_secondary = secondary_step(X, y, original_X, current_predictions, asset, target_returns, configs, reporting)
         
         # 4. Save the models
         reporting.all_assets.append(Reporting.Asset(ticker=asset[1], primary=training_step_primary, secondary=training_step_secondary))

training/meta_labeling.py

@@ -10,7 +10,7 @@ from reporting.types import Reporting
 
 def train_meta_labeling_model(
                             target_asset: str,
-                            X_pca: pd.DataFrame,
+                            X: pd.DataFrame,
                             input_predictions: pd.Series,
                             y: pd.Series,
                             target_returns: pd.Series,
@@ -28,9 +28,9 @@ def train_meta_labeling_model(
 
     print("Feature Selection started")
     backup_model = default_feature_selector_regression if data_config['method'] == 'regression' else default_feature_selector_classification
-    meta_feature_selection_input_X, meta_feature_selection_input_y = drop_until_first_valid_index(X_pca, meta_y)
+    meta_feature_selection_input_X, meta_feature_selection_input_y = drop_until_first_valid_index(X, meta_y)
     feature_selection_output = select_features(X = meta_feature_selection_input_X, y = meta_feature_selection_input_y, model = models[0][1], n_features_to_select = training_config['n_features_to_select'], backup_model = backup_model, scaling = training_config['scaler'])
-    meta_selected_features_X = X_pca[feature_selection_output.columns]
+    meta_selected_features_X = X[feature_selection_output.columns]
 
     meta_X = pd.concat([meta_selected_features_X, input_predictions, discretized_predictions], axis = 1)
 

training/training_steps.py

@@ -11,7 +11,6 @@ def primary_step(
                 X: pd.DataFrame,
                 y:pd.Series,
                 original_X:pd.DataFrame,
-                X_pca:pd.DataFrame,
                 asset:list,
                 target_returns:pd.Series,
                 configs: dict,
@@ -46,7 +45,7 @@ def primary_step(
             primary_model_predictions = current_predictions[model_name]
             primary_meta_result, primary_meta_preds, primary_meta_probabilities, meta_labeling_models = train_meta_labeling_model(
                 target_asset=asset[1],
-                X_pca = X_pca,
+                X = original_X,
                 input_predictions= primary_model_predictions,
                 y = y,
                 target_returns = target_returns,
@@ -73,7 +72,6 @@ def secondary_step(
                 X:pd.DataFrame,
                 y:pd.Series,
                 original_X:pd.DataFrame,
-                X_pca:pd.DataFrame,
                 current_predictions:pd.DataFrame,
                 asset:list,
                 target_returns:pd.Series,
@@ -114,7 +112,7 @@ def secondary_step(
             # 3. Train a Meta-labeling model on the averaged level-1 model predictions
             ensemble_meta_result, ensemble_meta_predictions, ensemble_meta_probabilities, ensemble_meta_labeling_models = train_meta_labeling_model(
                 target_asset=asset[1],
-                X_pca = X_pca,
+                X = original_X,
                 input_predictions= ensemble_predictions,
                 y = y,
                 target_returns = target_returns,