app.py

import streamlit as st
import pandas as pd
import joblib

# Load the model using joblib
model_path = "best_gb_model.joblib"  # Change to .joblib extension
model = joblib.load(model_path)

st.title("Loan Default Prediction")

st.write("""
This app predicts the likelihood of a loan default based on your input data. 
Please fill out the fields below to get a prediction.
""")

col1, col2, col3 = st.columns(3)

with col1:
    member_id = st.number_input('Member ID (int):', min_value=1, help="Unique identifier for the borrower. (Not used in prediction)")

with col2:
    total_payment = st.number_input('Total Payment (float):', min_value=0.0, help="Total amount paid by the borrower.")

with col3:
    income_to_loan_ratio = st.number_input('Income-to-Loan Ratio (float):', min_value=0.0, help="Ratio of the borrower\'s income to the loan amount.")

col4, col5, col6 = st.columns(3)

with col4:
    dti_revol_util = st.number_input('Debt-to-Income Ratio (float):', min_value=0.0, help="Debt-to-income ratio and revolving utilization.")

with col5:
    total_recovery = st.number_input('Total Recovery (float):', min_value=0.0, help="Total amount recovered from the borrower.")

with col6:
    balance_to_credit_ratio = st.number_input('Balance-to-Credit Ratio (float):', min_value=0.0, help="Ratio of the balance to the credit limit.")

col7, col8, col9 = st.columns(3)

with col7:
    recoveries_to_balance_ratio = st.number_input('Recoveries to Balance Ratio (float):', min_value=0.0, help="Ratio of recoveries to balance.")

with col8:
    batch_enrolled_to_total_rec_int = st.number_input('Batch Enrolled to Total Recovery Interest Ratio (float):', min_value=0.0, help="Ratio of batch enrolled to total recovery interest.")

with col9:
    loan_amnt_total_rec_int_ratio = st.number_input('Loan Amount to Total Recovery Interest Ratio (float):', min_value=0.0, help="Ratio of loan amount to total recovery interest.")

emp_length_missing = st.number_input('Employment Length Missing (int, 0 or 1):', min_value=0, max_value=1, help="Indicates if the employment length is missing (0 or 1).")

input_data = pd.DataFrame({
    'total_payment': [total_payment],
    'income_to_loan_ratio': [income_to_loan_ratio],
    'dti_revol_util': [dti_revol_util],
    'total_recovery': [total_recovery],
    'balance_to_credit_ratio': [balance_to_credit_ratio],
    'recoveries_to_balance_ratio': [recoveries_to_balance_ratio],
    'batch_enrolled_to_total_rec_int': [batch_enrolled_to_total_rec_int],
    'loan_amnt_total_rec_int_ratio': [loan_amnt_total_rec_int_ratio],
    'emp_length_missing': [emp_length_missing]
})

st.write('Selected input data (excluding Member ID):')
st.write(input_data)

if st.button('Predict'):
    try:
        prediction = model.predict(input_data)
        prediction_proba = model.predict_proba(input_data)[:, 1]  
        
        st.write(f'The predicted loan default status is: {"Default" if prediction[0] == 1 else "No Default"}')
        st.write(f'Probability of Default: {prediction_proba[0]:.2f}')

        if prediction[0] == 1:
            st.error('The model predicts that this loan is likely to default.')
        else:
            st.success('The model predicts that this loan is not likely to default.')
    except Exception as e:
        st.error(f"Error in prediction: {e}")