import gradio as gr
import random
import numpy as np
import pandas as pd
import joblib

# Define the function before loading the pipeline
def replace_binary_values(df):
    df = df.copy()
    df['Gender'] = df['Gender'].replace({'F': 1, 'M': 0})
    return df

# Now load the pre-trained pipeline
model_pipeline = joblib.load("full_model_pipeline.pkl")

# Define ranges and categories for input fields
numerical_ranges = {
    'Customer_Age': (26, 73),
    'Dependent_count': (0, 5),
    'Months_on_book': (13, 56),
    'Total_Relationship_Count': (1, 6),
    'Months_Inactive_12_mon': (0, 6),
    'Contacts_Count_12_mon': (0, 6),
    'Credit_Limit': (1438.3, 34516.0),
    'Total_Revolving_Bal': (0, 2517),
    'Avg_Open_To_Buy': (3.0, 34516.0),
    'Total_Amt_Chng_Q4_Q1': (0.0, 3.397),
    'Total_Trans_Amt': (510, 18484),
    'Total_Trans_Ct': (10, 139),
    'Total_Ct_Chng_Q4_Q1': (0.0, 3.714),
    'Avg_Utilization_Ratio': (0.0, 0.999)
}

categorical_options = {
    'Gender': ['M', 'F'],
    'Education_Level': ['High School', 'Graduate', 'Uneducated', 'Unknown', 'College', 'Post-Graduate', 'Doctorate'],
    'Marital_Status': ['Married', 'Single', 'Unknown', 'Divorced'],
    'Income_Category': ['$60K - $80K', 'Less than $40K', '$80K - $120K', '$40K - $60K', '$120K +', 'Unknown'],
    'Card_Category': ['Blue', 'Gold', 'Silver', 'Platinum']
}

# Function to auto-generate CLIENTNUM within its range
def generate_clientnum():
    return random.randint(708082083, 828343083)

# Define the prediction function using the loaded model pipeline
def predict_churn(
    Customer_Age, Gender, Dependent_count, Education_Level, Marital_Status, Income_Category, Card_Category,
    Months_on_book, Total_Relationship_Count, Months_Inactive_12_mon, Contacts_Count_12_mon,
    Credit_Limit, Total_Revolving_Bal, Avg_Open_To_Buy, Total_Amt_Chng_Q4_Q1,
    Total_Trans_Amt, Total_Trans_Ct, Total_Ct_Chng_Q4_Q1, Avg_Utilization_Ratio
):
    # Generate CLIENTNUM internally
    client_num = generate_clientnum()
    
    # Prepare input data for the model
    input_data = {
        "CLIENTNUM": client_num,  # Include CLIENTNUM as required by the model
        "Customer_Age": Customer_Age,
        "Gender": Gender,
        "Dependent_count": Dependent_count,
        "Education_Level": Education_Level,
        "Marital_Status": Marital_Status,
        "Income_Category": Income_Category,
        "Card_Category": Card_Category,
        "Months_on_book": Months_on_book,
        "Total_Relationship_Count": Total_Relationship_Count,
        "Months_Inactive_12_mon": Months_Inactive_12_mon,
        "Contacts_Count_12_mon": Contacts_Count_12_mon,
        "Credit_Limit": Credit_Limit,
        "Total_Revolving_Bal": Total_Revolving_Bal,
        "Avg_Open_To_Buy": Avg_Open_To_Buy,
        "Total_Amt_Chng_Q4_Q1": Total_Amt_Chng_Q4_Q1,
        "Total_Trans_Amt": Total_Trans_Amt,
        "Total_Trans_Ct": Total_Trans_Ct,
        "Total_Ct_Chng_Q4_Q1": Total_Ct_Chng_Q4_Q1,
        "Avg_Utilization_Ratio": Avg_Utilization_Ratio
    }
    input_df = pd.DataFrame([input_data])
    
    # Apply the binary transformation on Gender manually
    input_df = replace_binary_values(input_df)
    
    # Make prediction
    prediction = model_pipeline.predict(input_df)[0]
    
    # Interpret the prediction
    if prediction == 'Existing Customer':
        churn_status = "Tidak Churn"
    else:
        churn_status = "Churn"

    
    return churn_status

# Create Gradio input components for the app
inputs = [
    # Removed CLIENTNUM input from user inputs
    gr.Slider(numerical_ranges['Customer_Age'][0], numerical_ranges['Customer_Age'][1], step=1, label="Umur Nasabah"),
    gr.Dropdown(categorical_options['Gender'], label="Gender Nasabah"),
    gr.Slider(numerical_ranges['Dependent_count'][0], numerical_ranges['Dependent_count'][1], step=1, label="Jumlah Tanggungan"),
    gr.Dropdown(categorical_options['Education_Level'], label="Tingkat Pendidikan"),
    gr.Dropdown(categorical_options['Marital_Status'], label="Status Pernikahan"),
    gr.Dropdown(categorical_options['Income_Category'], label="Kategori Pemasukan"),
    gr.Dropdown(categorical_options['Card_Category'], label="Kategori Kartu"),
    gr.Slider(numerical_ranges['Months_on_book'][0], numerical_ranges['Months_on_book'][1], step=1, label="Lama Menjadi Nasabah (bulan)"),
    gr.Slider(numerical_ranges['Total_Relationship_Count'][0], numerical_ranges['Total_Relationship_Count'][1], step=1, label="Jumlah Hubungan Finansial"),
    gr.Slider(numerical_ranges['Months_Inactive_12_mon'][0], numerical_ranges['Months_Inactive_12_mon'][1], step=1, label="Bulan Tidak Aktif (12 bulan terakhir)"),
    gr.Slider(numerical_ranges['Contacts_Count_12_mon'][0], numerical_ranges['Contacts_Count_12_mon'][1], step=1, label="Jumlah Kontak (12 bulan terakhir)"),
    gr.Slider(numerical_ranges['Credit_Limit'][0], numerical_ranges['Credit_Limit'][1], step=100, label="Limit Kredit"),
    gr.Slider(numerical_ranges['Total_Revolving_Bal'][0], numerical_ranges['Total_Revolving_Bal'][1], step=10, label="Saldo Revolving"),
    gr.Slider(numerical_ranges['Avg_Open_To_Buy'][0], numerical_ranges['Avg_Open_To_Buy'][1], step=100, label="Sisa Limit Kredit"),
    gr.Slider(numerical_ranges['Total_Amt_Chng_Q4_Q1'][0], numerical_ranges['Total_Amt_Chng_Q4_Q1'][1], step=0.1, label="Perubahan Jumlah Transaksi (Q4 ke Q1)"),
    gr.Slider(numerical_ranges['Total_Trans_Amt'][0], numerical_ranges['Total_Trans_Amt'][1], step=100, label="Total Nilai Transaksi"),
    gr.Slider(numerical_ranges['Total_Trans_Ct'][0], numerical_ranges['Total_Trans_Ct'][1], step=1, label="Jumlah Total Transaksi"),
    gr.Slider(numerical_ranges['Total_Ct_Chng_Q4_Q1'][0], numerical_ranges['Total_Ct_Chng_Q4_Q1'][1], step=0.1, label="Perubahan Jumlah Transaksi (Q4 ke Q1)"),
    gr.Slider(numerical_ranges['Avg_Utilization_Ratio'][0], numerical_ranges['Avg_Utilization_Ratio'][1], step=0.01, label="Rata-rata Penggunaan Kredit"),
]

# Output component
output = gr.Textbox(label="Churn Status")

# Define Gradio app interface
app = gr.Interface(
    fn=predict_churn,
    inputs=inputs,
    outputs=output,
    title="Prediksi Churn Nasabah",
    description="Prediksi apakah seorang nasabah cenderung churn berdasarkan data yang diberikan."
)

# Launch the app
app.launch(share=True)