import streamlit as st
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import joblib
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, accuracy_score
import streamlit_authenticator as stauth


MODEL_FILENAME = "/tmp/insurance_churn_model.pkl"
st.title("Insurance Churn Prediction App")

menu = st.sidebar.radio("Navigation", ["Predict Churn","Train Model"])

if menu == "Train Model":
    # st.header("Upload Dataset and Train Model")
    # uploaded_file = st.file_uploader("Upload Insurance Churn Dataset (CSV)", type=["csv"])
    # if uploaded_file is not None:
    data = pd.read_csv("src/insurance.csv")
    st.subheader("Dataset Preview")
    st.dataframe(data.head())

    st.subheader("Summary Statistics")
    st.write(data.describe())

    if 'churn' in data.columns:
        st.subheader("Churn Distribution")
        fig, ax = plt.subplots()
        sns.countplot(x='churn', data=data, ax=ax)
        st.pyplot(fig)

    st.subheader("Model Training")
    target_column = st.selectbox("Select Target Column", options=data.columns, index=data.columns.get_loc('churn') if 'churn' in data.columns else 0)
    feature_columns = st.multiselect("Select Feature Columns", options=[col for col in data.columns if col != target_column])

    #if feature_columns and target_column:
        # X = pd.get_dummies(data[feature_columns])
        # y = data[target_column]

    # input features
    # Automatically exclude identifier columns
    # exclude_columns = ['Customer', 'Policy', 'Policy Number', 'Response']
    # feature_columns = [col for col in data.columns if col not in exclude_columns]
    # target_column = 'Response'

    # feature_columns = data.drop(columns=[target_column]).select_dtypes(include=[np.number]).columns.tolist()
    X= data[feature_columns]
    y = data[target_column]


    # X = data.drop('Response', axis = 1)
    # # output labels
    # y = data['Response']

    agree = st.checkbox("Continue Training")
    if agree:
        st.write("Great! ML Model Training Started..")
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

        model = RandomForestClassifier()
        model.fit(X_train, y_train)

        y_pred = model.predict(X_test)
        st.subheader("Model Performance")
        st.write("Accuracy:", accuracy_score(y_test, y_pred))
        st.text("Classification Report:")
        st.text(classification_report(y_test, y_pred))

        joblib.dump((model, X.columns.tolist()), MODEL_FILENAME)
        st.success(f"Model trained and saved as {MODEL_FILENAME}")

elif menu == "Predict Churn":
    st.header("Insurance Churn Predictor")
    st.markdown("To use Predictor, Please Train the ML Model if not done yet! ")
    

    try:
        model, feature_names = joblib.load(MODEL_FILENAME)
        st.success("Model loaded successfully.")
    except:
        st.error("Model not found. Please train the model first.")
        st.stop()

    st.subheader("Enter Customer Details")
    input_data = {}
    for feature in feature_names:
        input_data[feature] = st.text_input(f"{feature}", "")

    if st.button("Predict Churn"):
        try:
            input_df = pd.DataFrame([input_data])
            input_df = pd.get_dummies(input_df)

            for col in feature_names:
                if col not in input_df.columns:
                    input_df[col] = 0
            input_df = input_df[feature_names]

            prediction = model.predict(input_df)[0]
            st.subheader("Prediction Result")
            st.write(f"Churn: {'Yes' if prediction == 1 else 'No'}")
        except Exception as e:
            st.error(f"Error in prediction: {e}")