Update superkart_sales_forecast.pkl

superkart_sales_forecast.pkl

@@ -1,26 +1,48 @@
+# model_export.py
+
 import joblib
-from sklearn.ensemble import RandomForestRegressor  # or your actual model
+import pandas as pd
+from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor  # or whatever models you trained
+
+# Replace with your actual evaluation function
+def evaluate_model(y_true, y_pred):
+    from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
+    return {
+        "RMSE": mean_squared_error(y_true, y_pred, squared=False),
+        "MAE": mean_absolute_error(y_true, y_pred),
+        "R2": r2_score(y_true, y_pred)
+    }
+
+# Assume rf_best_model and gb_best_model are already trained
+# Also assume we have performance_df with RMSE values
 
+# Example: Simulate performance_df (you should use the real one)
+# import pandas as pd
+# performance_df = pd.DataFrame({
+#     "RMSE": {
+#         "Random Forest (Tuned)": 113.52,
+#         "Gradient Boosting (Tuned)": 119.40
+#     }
+# })
 
 # -----------------------------
 # Select Final Model Based on RMSE
 # -----------------------------
-# 'rf_best_model' and 'gb_best_model' are tuned models (Random Forest, Gradient Boosting)
 best_model = rf_best_model if performance_df["RMSE"].idxmin() == "Random Forest (Tuned)" else gb_best_model
 
-# Make predictions and evaluate
+# Predict and evaluate
 final_predictions = best_model.predict(X_test)
 final_metrics = evaluate_model(y_test, final_predictions)
 
-print("\n Best Model Selected:", performance_df['RMSE'].idxmin())
-print(" Performance on Test Set:", final_metrics)
+print("\nBest Model Selected:", performance_df['RMSE'].idxmin())
+print("Performance on Test Set:", final_metrics)
 
 # -----------------------------
 # Serialize the Best Model
 # -----------------------------
-model_filename = "superkart_sales_forecast.pkl"  # Hugging Face expects this name
+model_filename = "superkart_sales_forecast.pkl"
 joblib.dump(best_model, model_filename)
-print(f"\n Model serialized and saved as: {model_filename}")
+print(f"\nModel serialized and saved as: {model_filename}")
 
 # -----------------------------
 # Load and Validate the Model
@@ -29,4 +51,4 @@ loaded_model = joblib.load(model_filename)
 loaded_predictions = loaded_model.predict(X_test)
 loaded_metrics = evaluate_model(y_test, loaded_predictions)
 
-print("\n Loaded Model Performance on Test Set:", loaded_metrics)
+print("\nLoaded Model Performance on Test Set:", loaded_metrics)