Update serve_model.py

flask-app/serve_model.py

@@ -1,5 +1,8 @@
 from flask import Flask, request, jsonify
 import pandas as pd
+from lime import lime_tabular
+import shap
+import lime
 import pickle
 import os
 import sys
@@ -17,6 +20,7 @@
 def home():
     return ("Welcome to ML based fraud detction")
 
+
 @app.route('/predict', methods=['POST'])
 def detect():
     try:
@@ -27,21 +31,29 @@ def detect():
         df['signup_time'] = pd.to_datetime(df['signup_time'])
         df['purchase_time'] = pd.to_datetime(df['purchase_time'])
 
-        #  FeatureEngineering
+        # Feature Engineering
         fg = FeatureEngineering(df)
         featured_df = fg.preprocess()
         if len(featured_df) > 0:
             # Make detection
             result = model.predict(featured_df.values).tolist()
             detection_results = ["Fraud" if res == 1 else "Not Fraud" for res in result]
-            return jsonify({'Detection': detection_results})
+
+            # SHAP Explanation
+            explainer = shap.TreeExplainer(model)
+            shap_values = explainer.shap_values(featured_df)
+            shap_summary = shap.summary_plot(shap_values, featured_df, plot_type="bar")
+
+            # LIME Explanation
+            lime_explainer = lime_tabular.LimeTabularExplainer(featured_df.values, feature_names=featured_df.columns)
+            lime_exp = lime_explainer.explain_instance(featured_df.iloc[0].values, model.predict_proba, num_features=5)
+            lime_explanation = lime_exp.as_html()
+
+            return jsonify({'Detection': detection_results, 'SHAP_Explanation': shap_summary, 'LIME_Explanation': lime_explanation})
         else:
             return jsonify({'error': 'No data provided for detection'}), 400
 
     except KeyError as ke:
         return jsonify({'error': f"Missing key in the input data: {ke}"}), 400
     except Exception as e:
         return jsonify({'error': str(e)}), 400
-
-if __name__ == '__main__':
-    app.run(debug=True)