Update app.py

app.py

@@ -1,25 +1,47 @@
 import streamlit as st
-from transformers import pipeline
+import pandas as pd
+import joblib
 
-# Load pretrained fraud detection model from Hugging Face
-@st.cache_resource
-def load_model():
-    return pipeline("text-classification", model="d4data/bert-base-cased-finetuned-fraud-detection")
+# Load the trained model and preprocessing objects
+model = joblib.load("fraud_detection_model.joblib")
+label_encoders = joblib.load("label_encoders.joblib")
+scaler = joblib.load("scaler.joblib")
 
-model = load_model()
+# Streamlit app
+st.title("Credit Card Fraud Detection")
 
-# Streamlit UI
-st.title("💳 Financial Fraud Detection System (Hugging Face Model)")
+# Input fields for user
+st.header("Enter Transaction Details")
+amount = st.number_input("Amount", min_value=0.0)
+merchant_id = st.text_input("Merchant ID")
+transaction_type = st.selectbox("Transaction Type", ["purchase", "refund"])
+location = st.text_input("Location")
 
-# User input text
-user_input = st.text_area("Enter transaction description:", "Payment of $500 for electronics purchase")
+# Preprocess input data
+if st.button("Predict"):
+    # Create a DataFrame from the input
+    input_data = pd.DataFrame({
+        "Amount": [amount],
+        "MerchantID": [merchant_id],
+        "TransactionType": [transaction_type],
+        "Location": [location]
+    })
 
-if st.button("Check for Fraud"):
-    prediction = model(user_input)
-    label = prediction[0]["label"]
-    score = prediction[0]["score"]
+    # Apply label encoding to categorical columns
+    for col, le in label_encoders.items():
+        input_data[col] = le.transform(input_data[col])
 
-    if label == "LABEL_1":  # Assuming LABEL_1 means fraud
-        st.error(f"🚨 Fraudulent Transaction Detected! (Confidence: {score:.2f})")
+    # Scale the "Amount" column
+    input_data["Amount"] = scaler.transform(input_data[["Amount"]])
+
+    # Make prediction
+    prediction = model.predict(input_data)
+    prediction_proba = model.predict_proba(input_data)
+
+    # Display the result
+    if prediction[0] == 1:
+        st.error("Fraudulent Transaction Detected!")
     else:
-        st.success(f"✅ Legitimate Transaction (Confidence: {score:.2f})")
+        st.success("Legitimate Transaction")
+
+    st.write(f"Probability: {prediction_proba[0][1]:.2f}")