Update app.py

app.py

@@ -1,14 +1,18 @@
 import gradio as gr
 import pandas as pd
 import pickle
+import shap
 
-# Load the saved model and feature columns
+# Load the saved model, feature columns, and SHAP explainer
 with open('logistic_regression_model.pkl', 'rb') as model_file:
     model = pickle.load(model_file)
 
 with open('feature_columns.pkl', 'rb') as col_file:
     feature_columns = pickle.load(col_file)
 
+# Create a SHAP explainer
+explainer = shap.Explainer(model)
+
 def predict_status(UDI, air_temp, proc_temp, speed, torque, tool_wear):
     # Prepare the input data
     sample_input = {
@@ -36,10 +40,19 @@ def predict_status(UDI, air_temp, proc_temp, speed, torque, tool_wear):
 
     # Make a prediction
     prediction = model.predict(sample_df)[0]
-
-    # Return the result based on the prediction
+    
+    # Get SHAP values
+    shap_values = explainer(sample_df)
+    
+    # Determine the most impactful features if prediction is 0 (machine is not working)
     if prediction == 0:
-        return "🔴 Machine is NOT working."
+        explanation = shap_values[0].values
+        features = list(sample_df.columns)
+        feature_importance = sorted(zip(features, explanation), key=lambda x: abs(x[1]), reverse=True)
+        reason = "Machine is NOT working due to the following features:\n"
+        for feature, importance in feature_importance:
+            reason += f"{feature}: {'High' if importance > 0 else 'Low'} impact\n"
+        return reason
     else:
         return "🟢 Machine is working."