import gradio as gr
import pandas as pd
import joblib
import os

# Load the models
preprocessor_path = os.path.join('.', 'preprocessor.joblib')
model_path = os.path.join('.', 'Best_model.joblib')

preprocessor = joblib.load(preprocessor_path)
best_model = joblib.load(model_path)

# Define prediction function
def predict(age, job, marital, education, default, housing, loan, contact, month, day_of_week, duration, campaign, pdays, previous, poutcome, emp_var_rate, cons_price_idx, cons_conf_idx, euribor3m, nr_employed):
    # Create a DataFrame for the input data
    data = pd.DataFrame({
        'age': [age],
        'job': [job],
        'marital': [marital],
        'education': [education],
        'default': [default],
        'housing': [housing],
        'loan': [loan],
        'contact': [contact],
        'month': [month],
        'day_of_week': [day_of_week],
        'duration': [duration],
        'campaign': [campaign],
        'pdays': [pdays],
        'previous': [previous],
        'poutcome': [poutcome],
        'emp.var.rate': [emp_var_rate],
        'cons.price.idx': [cons_price_idx],
        'cons.conf.idx': [cons_conf_idx],
        'euribor3m': [euribor3m],
        'nr.employed': [nr_employed]
    })

    # Preprocess the data
    preprocessed_data = preprocessor.transform(data)

    # Make predictions
    prediction = best_model.predict(preprocessed_data)
    probability = best_model.predict_proba(preprocessed_data)

    return {
        "Prediction": prediction[0],
        "Probability (Yes)": probability[0][1],
        "Probability (No)": probability[0][0]
    }

# Define the interface
def gradio_interface():
    with gr.Blocks() as app:
        gr.Markdown("# Term Deposit Subscription Prediction")

        with gr.Row():
            age = gr.Number(label="Age", value=30)
            job = gr.Dropdown(["housemaid", "services", "admin.", "blue-collar", "technician", "retired", "management", "unemployed", "self-employed", "unknown", "entrepreneur", "student"], label="Job")
            marital = gr.Dropdown(["married", "single", "divorced", "unknown"], label="Marital Status")
            education = gr.Dropdown(["basic.4y", "high.school", "basic.6y", "basic.9y", "professional.course", "unknown", "university.degree", "illiterate", "tertiary", "secondary", "primary"], label="Education")

        with gr.Row():
            default = gr.Dropdown(["no", "unknown", "yes"], label="Default")
            housing = gr.Dropdown(["no", "yes", "unknown"], label="Housing Loan")
            loan = gr.Dropdown(["no", "yes", "unknown"], label="Personal Loan")
            contact = gr.Dropdown(["telephone", "cellular", "unknown"], label="Contact Type")

        with gr.Row():
            month = gr.Dropdown(["may", "jun", "jul", "aug", "oct", "nov", "dec", "mar", "apr", "sep", "jan", "feb"], label="Month")
            day_of_week = gr.Dropdown(["mon", "tue", "wed", "thu", "fri"], label="Day of Week")

        with gr.Row():
            duration = gr.Number(label="Call Duration (seconds)", value=100)
            campaign = gr.Number(label="Number of Contacts during Campaign", value=1)
            pdays = gr.Number(label="Days since Last Contact", value=999)
            previous = gr.Number(label="Number of Contacts before Campaign", value=0)
            poutcome = gr.Dropdown(["nonexistent", "failure", "success", "unknown", "other"], label="Previous Outcome")

        with gr.Row():
            emp_var_rate = gr.Number(label="Employment Variation Rate", value=1.1)
            cons_price_idx = gr.Number(label="Consumer Price Index", value=93.994)
            cons_conf_idx = gr.Number(label="Consumer Confidence Index", value=-36.4)
            euribor3m = gr.Number(label="Euribor 3-Month Rate", value=4.857)
            nr_employed = gr.Number(label="Number of Employees", value=5191.0)

        predict_btn = gr.Button("Predict")
        output = gr.JSON()

        predict_btn.click(
            predict, 
            inputs=[age, job, marital, education, default, housing, loan, contact, month, day_of_week, duration, campaign, pdays, previous, poutcome, emp_var_rate, cons_price_idx, cons_conf_idx, euribor3m, nr_employed], 
            outputs=output
        )

    return app

# Launch the app
if __name__ == "__main__":
    app = gradio_interface()
    app.launch()