app.py

import streamlit as st
import pandas as pd
import numpy as np
import tensorflow as tf
import joblib

# Load trained model
model = tf.keras.models.load_model("banking_model.keras")

# Load encoders and scaler
label_encoders = joblib.load("label_encoders.pkl")
scaler = joblib.load("scaler.pkl")

# Define feature names
numerical_features = ["DPD", "Credit Expiration"]
binary_features = ["Feature1", "Feature2", "Feature3"]  # Replace with actual binary features
stage_feature = "Stage As Last Month"

st.title("Classification Prediction App")

# Create input fields for user input
user_input = {}

# Numerical inputs (DPD, Credit Expiration)
for feature in numerical_features:
    user_input[feature] = st.number_input(f"Enter {feature}", value=0, min_value=0)

# Binary features (Yes/No)
for feature in binary_features:
    user_input[feature] = st.selectbox(f"{feature} (Yes/No)", ["Yes", "No"])
    user_input[feature] = 1 if user_input[feature] == "Yes" else 0  # Convert to 1/0

# Stage as Last Month (Dropdown 1, 2, 3)
user_input[stage_feature] = st.selectbox("Stage As Last Month", [1, 2, 3])

# Convert input to DataFrame
input_df = pd.DataFrame([user_input])

# Apply scaling
input_df[numerical_features] = scaler.transform(input_df[numerical_features])

# Predict when user clicks button
if st.button("Predict"):
    prediction = model.predict(input_df)
    predicted_stage = np.argmax(prediction)
    st.success(f"Predicted Stage: {predicted_stage}")



if __name__ == "__main__":
    main()