Fix: Pin scikit-learn version and add powerbi_export

powerbi_export.py

@@ -0,0 +1,59 @@
+"""
+Power BI export module for fraud detection pipeline.
+Exports transaction data and metrics in Power BI-friendly format.
+"""
+
+import pandas as pd
+from pathlib import Path
+import sys
+import os
+
+# Add powerbi directory to path to import prepare_powerbi_data
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+try:
+    from powerbi.prepare_powerbi_data import export_for_powerbi, POWERBI_OUT
+except ImportError:
+    # Fallback if powerbi module structure is different
+    POWERBI_OUT = Path("powerbi/out")
+    POWERBI_OUT.mkdir(parents=True, exist_ok=True)
+
+
+def export_powerbi_csvs(result_df: pd.DataFrame, output_dir: Path = None):
+    """
+    Export fraud detection results to Power BI format.
+    
+    This function is called by the Gradio app to automatically export
+    scored transactions and metrics for Power BI consumption.
+    
+    Args:
+        result_df: DataFrame with fraud detection results (must include
+                  'fraud_probability', 'fraud_prediction', 'risk_level')
+        output_dir: Optional output directory (defaults to powerbi/out)
+    
+    Returns:
+        tuple: (transactions_path, metrics_path)
+    """
+    if output_dir is None:
+        output_dir = POWERBI_OUT
+    
+    # Ensure output directory exists
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    
+    try:
+        # Use the existing export function from powerbi module
+        tx_path, metrics_path = export_for_powerbi(result_df, output_dir)
+        return tx_path, metrics_path
+    except Exception as e:
+        # If the powerbi module isn't available, do a simple export
+        print(f"⚠️ Warning: Could not use full Power BI export: {e}")
+        print("   Falling back to simple CSV export...")
+        
+        # Simple fallback export
+        tx_path = output_dir / "transactions_scored.csv"
+        result_df.to_csv(tx_path, index=False)
+        print(f"✅ Exported {len(result_df)} transactions to {tx_path}")
+        
+        return tx_path, None
+

requirements.txt

@@ -1,7 +1,7 @@
 # Core ML and Data Processing
 pandas>=2.0.0,<3.0.0
 numpy>=1.26.0,<2.0.0
-scikit-learn>=1.4.0,<2.0.0
+scikit-learn==1.7.1
 joblib>=1.3.0,<2.0.0
 
 # Gradio and Web Interface