Update app.py

app.py

@@ -54,16 +54,20 @@ for col in required_columns:
     if col not in df.columns:
         raise Exception(f"❌ Column '{col}' is missing from the dataset!")
 
-# ✅ Convert "Case Problem" column to lowercase and remove spaces for consistency
+# ✅ Convert "Case Problem" & "Feedback" to lowercase and remove spaces
 df["Case Problem"] = df["Case Problem"].astype(str).str.strip().str.lower()
+df["Feedback"] = df["Feedback"].astype(str).str.strip().str.lower()
 
-# ✅ Train and save LabelEncoder
-encoder = LabelEncoder()
-df["Case Problem Encoded"] = encoder.fit_transform(df["Case Problem"])
-df["Feedback Encoded"] = encoder.fit_transform(df["Feedback"])
+# ✅ Train and save LabelEncoders for both input and output
+case_problem_encoder = LabelEncoder()
+feedback_encoder = LabelEncoder()
 
-# ✅ Save encoder
-joblib.dump(encoder, "case_problem_encoder.pkl")
+df["Case Problem Encoded"] = case_problem_encoder.fit_transform(df["Case Problem"])
+df["Feedback Encoded"] = feedback_encoder.fit_transform(df["Feedback"])
+
+# ✅ Save encoders
+joblib.dump(case_problem_encoder, "case_problem_encoder.pkl")
+joblib.dump(feedback_encoder, "feedback_encoder.pkl")
 
 # ✅ Train Model
 X = df[["Case Problem Encoded"]]
@@ -86,8 +90,9 @@ async def predict_feedback(data: PredictionInput):
     if model is None:
         return {"error": "Model is not trained yet."}
 
-    # ✅ Load encoder
-    encoder = joblib.load("case_problem_encoder.pkl")
+    # ✅ Load encoders
+    case_problem_encoder = joblib.load("case_problem_encoder.pkl")
+    feedback_encoder = joblib.load("feedback_encoder.pkl")
 
     # ✅ Convert input to lowercase and remove spaces
     case_problem_lower = data.case_problem.strip().lower()
@@ -101,9 +106,9 @@ async def predict_feedback(data: PredictionInput):
         }
 
     try:
-        case_problem_encoded = encoder.transform([case_problem_lower])
+        case_problem_encoded = case_problem_encoder.transform([case_problem_lower])
         prediction = model.predict([[case_problem_encoded[0]]])
-        feedback_predicted = encoder.inverse_transform(prediction)[0]
+        feedback_predicted = feedback_encoder.inverse_transform(prediction)[0]
         return {"Predicted Feedback": feedback_predicted}
     except Exception as e:
         return {"error": str(e)}
@@ -113,7 +118,9 @@ def gradio_interface(case_problem):
     if model is None:
         return "Model not trained yet."
 
-    encoder = joblib.load("case_problem_encoder.pkl")
+    case_problem_encoder = joblib.load("case_problem_encoder.pkl")
+    feedback_encoder = joblib.load("feedback_encoder.pkl")
+
     case_problem_lower = case_problem.strip().lower()
 
     if case_problem_lower not in df["Case Problem"].values:
@@ -121,9 +128,9 @@ def gradio_interface(case_problem):
         return f"Invalid case problem. Please enter a valid category. Options: {valid_problems}"
 
     try:
-        case_problem_encoded = encoder.transform([case_problem_lower])
+        case_problem_encoded = case_problem_encoder.transform([case_problem_lower])
         prediction = model.predict([[case_problem_encoded[0]]])
-        feedback_predicted = encoder.inverse_transform(prediction)[0]
+        feedback_predicted = feedback_encoder.inverse_transform(prediction)[0]
         return f"Predicted Feedback: {feedback_predicted}"
     except Exception as e:
         return f"Error: {str(e)}"