Update app.py

app.py

@@ -1,29 +1,42 @@
 import gradio as gr
 import pickle
+import numpy as np
 from sklearn.preprocessing import StandardScaler
 
-# Load the model
+# Load the trained RandomForest model
 with open("fraud_detection_model.pkl", "rb") as f:
     model = pickle.load(f)
 
-# Define a function for prediction
+# Load the scaler used during training (ensure it's the same one used for training the model)
+with open("fraud_detection_model.pkl", "rb") as f:
+    scaler = pickle.load(f)
+
+# Define the prediction function
 def predict(input_data):
+    # Convert input data to a list of floats
     input_data = [float(x) for x in input_data.split(",")]
-    scaler = StandardScaler()
-    input_data = scaler.fit_transform([input_data])
+    
+    # Preprocess the input data (e.g., scaling)
+    input_data = np.array(input_data).reshape(1, -1)  # Reshape for a single sample
+    input_data = scaler.transform(input_data)  # Apply the same scaling as during training
+    
+    # Make the prediction using the loaded model
     prediction = model.predict(input_data)
+    
+    # Return the result as "Fraud" or "Not Fraud"
     return "Fraud" if prediction[0] == 1 else "Not Fraud"
 
-# Updated Gradio Interface
+# Gradio Interface
 input_text = gr.Textbox(label="Input Features (comma-separated)", placeholder="Enter features like 1.2, 3.4, ...")
 output_text = gr.Textbox(label="Prediction")
 
+# Create the Gradio interface
 interface = gr.Interface(
     fn=predict,
     inputs=input_text,
     outputs=output_text,
     title="Fraud Detection System",
-    description="Enter the features to predict whether it is fraud or not."
+    description="Enter the features in comma-separated format to predict whether it is fraud or not."
 )
 
 # Run the application