Update app.py

app.py

@@ -3,7 +3,7 @@ import pandas as pd
 import pickle
 import os
 
-# Helper function to load pickle files
+# Load pickle files
 def load_pickle_file(file_name):
     file_path = os.path.join(os.path.dirname(__file__), file_name)
     try:
@@ -15,19 +15,16 @@ def load_pickle_file(file_name):
     except Exception as e:
         return f"An error occurred while loading {file_name}: {e}"
 
-# Load the pre-trained model and label encoder
+# Load model and label encoder
 model = load_pickle_file('best_model.pkl')
 label_encoder = load_pickle_file('label_encoder.pkl')
 
-# Check if the model and label encoder were loaded successfully
 if isinstance(model, str) or isinstance(label_encoder, str):
-    # Print the error message for debugging
-    print(f"Error loading model or label encoder: {model} | {label_encoder}")
     raise Exception(f"Error loading model or label encoder: {model} | {label_encoder}")
 
-# Define the prediction function
+# Prediction function
 def predict_coffee_type(time_of_day, coffee_strength, sweetness_level, milk_type, coffee_temperature, flavored_coffee, caffeine_tolerance, coffee_bean, coffee_size, dietary_preferences):
-    # Input Data
+    # Prepare input data
     input_data = pd.DataFrame({
         'Token_0': [time_of_day],
         'Token_1': [coffee_strength],
@@ -41,21 +38,18 @@ def predict_coffee_type(time_of_day, coffee_strength, sweetness_level, milk_type
         'Token_9': [dietary_preferences]
     })
 
-    # One-hot encode the input data (ensure it matches the training data)
+    # One-hot encode the input data
     input_encoded = pd.get_dummies(input_data)
-    required_columns = model.feature_names_in_  # Ensure that the input has the correct columns
+    required_columns = model.feature_names_in_  # Ensure columns match training data
     for col in required_columns:
         if col not in input_encoded.columns:
-            input_encoded[col] = 0  # Add missing columns as 0
-    
-    input_encoded = input_encoded[required_columns]  # Ensure the order of columns matches the training data
-    
-    # Make prediction
+            input_encoded[col] = 0  # Add missing columns
+
+    input_encoded = input_encoded[required_columns]  # Reorder columns to match model training
+
+    # Predict the coffee type
     prediction = model.predict(input_encoded)[0]
-    
-    # Decode the label
     coffee_type = label_encoder.inverse_transform([prediction])[0]
-    
     return f"Recommended Coffee: {coffee_type}"
 
 # Set up Gradio interface