app.py

import streamlit as st
import pandas as pd
import numpy as np
import pickle
import os
from huggingface_hub import hf_hub_download
from sklearn.preprocessing import StandardScaler, LabelEncoder

# Hugging Face repo details
HF_REPO_ID = "wvsu-dti-aidev-team/customer_churn_logres_model"
MODEL_FILENAME = "customer_churn_logres_model.pkl"

# Download and load the trained model
st.write("## Telco Customer Churn Prediction")

hf_token = os.getenv("HF_TOKEN")
if not hf_token:
    st.error("HF_TOKEN environment variable not set. Please configure it before proceeding.")
else:
    with st.spinner("Downloading the model from Hugging Face..."):
        model_path = hf_hub_download(repo_id=HF_REPO_ID, filename=MODEL_FILENAME, token=hf_token)

    # Load the model
    with open(model_path, "rb") as f:
        model = pickle.load(f)
    
    st.success("Model loaded successfully!")

    # Define feature names (from dataset)
    feature_names = [
        "gender", "SeniorCitizen", "Partner", "Dependents", "tenure", 
        "PhoneService", "MultipleLines", "InternetService", "OnlineSecurity", 
        "OnlineBackup", "DeviceProtection", "TechSupport", "StreamingTV", 
        "StreamingMovies", "Contract", "PaperlessBilling", "PaymentMethod", 
        "MonthlyCharges", "TotalCharges"
    ]

    # Define categorical features for encoding
    categorical_features = ["gender", "InternetService", "Contract", "PaymentMethod"]

    # Create input fields for each feature
    st.write("### Enter Customer Details")

    user_input = {}
    for feature in feature_names:
        if feature in categorical_features:
            user_input[feature] = st.selectbox(f"{feature}:", ["DSL", "Fiber optic", "No"])
        elif feature in ["SeniorCitizen", "Partner", "Dependents", "PhoneService", "MultipleLines", 
                         "OnlineSecurity", "OnlineBackup", "DeviceProtection", "TechSupport", 
                         "StreamingTV", "StreamingMovies", "PaperlessBilling"]:
            user_input[feature] = st.radio(f"{feature}:", [0, 1])
        else:
            user_input[feature] = st.number_input(f"{feature}:", min_value=0.0, step=0.1)

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

    # Encode categorical features using LabelEncoder
    label_encoders = {}
    for feature in categorical_features:
        le = LabelEncoder()
        input_df[feature] = le.fit_transform(input_df[feature])
        label_encoders[feature] = le

    # Preprocess input: Apply scaling
    scaler = StandardScaler()
    input_scaled = scaler.fit_transform(input_df)

    # Predict churn
    if st.button("Predict Customer Churn"):
        prediction = model.predict(input_scaled)[0]
        st.write("## Prediction:")
        if prediction == 1:
            st.error("⚠️ This customer is likely to churn!")
        else:
            st.success("✅ This customer is likely to stay.")