app.py

import gradio as gr
from huggingface_hub import hf_hub_download
import joblib
import numpy as np
import os

# Download the model and scaler from your repository
model_path = hf_hub_download(repo_id="NiharMandahas/RF_Customer_Fraud", filename="random_forest_model.joblib")
scaler_path = hf_hub_download(repo_id="NiharMandahas/RF_Customer_Fraud", filename="rf_scaler.joblib")

# Load the model and scaler
model = joblib.load(model_path)
scaler = joblib.load(scaler_path)

def predict_fraud(account_age, cred_changes_freq, return_order_ratio, vpn_usage, credit_score):
    # Prepare input
    input_data = np.array([[
        account_age,
        cred_changes_freq,
        return_order_ratio,
        vpn_usage,
        credit_score
    ]])
    
    # Scale input
    scaled_input = scaler.transform(input_data)
    
    # Get prediction and probability
    prediction = model.predict(scaled_input)[0]
    probabilities = model.predict_proba(scaled_input)[0]
    
    # Prepare result
    result = "Fraud" if prediction == 1 else "Not Fraud"
    confidence = float(probabilities[prediction])
    
    return f"Prediction: {result}\nConfidence: {confidence:.2f}"

# Create Gradio interface
iface = gr.Interface(
    fn=predict_fraud,
    inputs=[
        gr.Number(label="Account Age (months)"),
        gr.Number(label="Credential Changes Frequency (per year)"),
        gr.Number(label="Return to Order Ratio"),
        gr.Number(label="VPN Usage (0 or 1)"),
        gr.Number(label="Credit Score")
    ],
    outputs="text",
    title="Fraud Detection Model",
    description="Enter account details to predict potential fraud"
)

iface.launch()