app.py

import gradio as gr
import pandas as pd
import numpy as np
import pickle
import os

# Path to your saved model artifact
MODEL_PATH = 'car_price_model.pkl'

# --- Load model from pickle --- #
if not os.path.exists(MODEL_PATH):
    print(f"Error: Model file not found at {MODEL_PATH}. Please ensure 'car_price_model.pkl' is in the same directory.")
    # In a real deployment, you might want to handle this more robustly, e.g., by exiting or raising an error.
    exit(1)

with open(MODEL_PATH, 'rb') as f:
    artifact = pickle.load(f)

pipeline   = artifact['pipeline']
brand_list = sorted(artifact['brand_list'])
fuel_list  = sorted(artifact['fuel_list'])
owner_list = ['First Owner', 'Second Owner', 'Third Owner', 'Fourth & Above Owner', 'Test Drive Car']


def predict_price(
    brand, km_driven, fuel, seller_type, transmission,
    owner, mileage, engine, max_power, seats, car_age
):
    try:
        input_df = pd.DataFrame([{ # Ensure correct feature names as expected by the pipeline
            'km_driven'   : float(km_driven),
            'fuel'        : fuel,
            'seller_type' : seller_type,
            'transmission': transmission,
            'owner'       : owner,
            'mileage'     : float(mileage),
            'engine'      : float(engine),
            'max_power'   : float(max_power),
            'seats'       : int(seats),
            'car_age'     : int(car_age),
            'brand'       : brand,
        }])

        log_pred = pipeline.predict(input_df)[0]
        price    = np.expm1(log_pred)

        # Confidence band ±12% (as per original notebook)
        low  = price * 0.88
        high = price * 1.12

        result = (
            f"🚗 Estimated Selling Price\n"
            f"{'─' * 35}\n"
            f"  💰 Predicted  : ₹ {price:>12,.0f}\n"
            f"  📉 Low Range  : ₹ {low:>12,.0f}\n"
            f"  📈 High Range : ₹ {high:>12,.0f}\n"
            f"{'─' * 35}\n"
            f"  Model : {artifact['model_name']}\n"
            f"  R²    : {artifact['test_r2']:.4f}"
        )
        return result

    except Exception as e:
        return f'⚠️ Error: {str(e)}'


# ── Gradio UI ──────────────────────────────────────────────
with gr.Blocks(
    title='🚗 Car Price Predictor',
    theme=gr.themes.Soft(primary_hue='blue')
) as demo:

    gr.Markdown("""
    # 🚗 Used Car Price Predictor
    ### Powered by Machine Learning · Dataset: Car Details v3
    Fill in the car details below and click **Predict Price** to get an instant estimate.
    """)

    with gr.Row():
        # ── Left Column: Car Identity
        with gr.Column(scale=1):
            gr.Markdown('### 🔖 Car Identity')
            brand        = gr.Dropdown(brand_list, label='Brand', value='Maruti')
            fuel         = gr.Radio(fuel_list, label='Fuel Type', value='Petrol')
            transmission = gr.Radio(['Manual', 'Automatic'], label='Transmission', value='Manual')
            seller_type  = gr.Radio(['Individual', 'Dealer', 'Trustmark Dealer'],
                                     label='Seller Type', value='Individual')
            owner        = gr.Dropdown(owner_list, label='Owner', value='First Owner')

        # ── Right Column: Specs
        with gr.Column(scale=1):
            gr.Markdown('### ⚙️ Car Specifications')
            car_age   = gr.Slider(1, 40, value=5, step=1, label='Car Age (years)')
            km_driven = gr.Number(value=50000, label='KM Driven')
            mileage   = gr.Slider(5.0, 35.0, value=18.0, step=0.1, label='Mileage (kmpl)')
            engine    = gr.Slider(500, 5000, value=1200, step=50, label='Engine (CC)')
            max_power = gr.Slider(30, 400, value=80, step=1, label='Max Power (bhp)')
            seats     = gr.Slider(2, 10, value=5, step=1, label='Seats')

    with gr.Row():
        predict_btn  = gr.Button('🚀 Predict Price', variant='primary', scale=2)
        clear_btn    = gr.Button('🔄 Reset', scale=1)

    output = gr.Textbox(label='Prediction Result', lines=9, )

    # Examples
    gr.Examples(
        examples=[
            ['Maruti',   50000,  'Petrol',  'Individual',      'Manual',    'First Owner',  22.0, 1200, 82,  5, 5],
            ['Honda',    30000,  'Petrol',  'Individual',      'Automatic', 'Second Owner', 17.0, 1500, 118, 5, 3],
            ['Toyota',   80000,  'Diesel',  'Dealer',          'Manual',    'First Owner',  21.0, 2000, 150, 7, 7],
            ['BMW',      40000,  'Petrol',  'Trustmark Dealer','Automatic', 'Second Owner', 12.0, 2000, 190, 5, 4],
            ['Mahindra', 120000, 'Diesel',  'Individual',      'Manual',    'Third Owner',  16.0, 2200, 130, 7, 10],
        ],
        inputs=[brand, km_driven, fuel, seller_type, transmission,
                owner, mileage, engine, max_power, seats, car_age],
        label='📋 Example Cars — Click to Auto-Fill'
    )

    gr.Markdown(f"""
    ---
    **Model Info:** {artifact['model_name']} · Test R² = {artifact['test_r2']:.4f} · MAE ≈ ₹{artifact['test_mae']:,.0f}
    """)

    predict_btn.click(
        fn=predict_price,
        inputs=[brand, km_driven, fuel, seller_type, transmission,
                owner, mileage, engine, max_power, seats, car_age],
        outputs=output
    )
    clear_btn.click(fn=lambda: None, outputs=output)


if __name__ == '__main__':
    demo.launch(share=False, show_error=True)