app.py

from __future__ import annotations

from pathlib import Path

import gradio as gr

from src.data import load_training_data
from src.modeling import DEFAULT_MODEL_PATH, HousePriceModel, train_model


MODEL_PATH = Path(DEFAULT_MODEL_PATH)


def ensure_model() -> HousePriceModel:
    if not MODEL_PATH.exists():
        train_model(load_training_data(), artifact_path=MODEL_PATH)
    return HousePriceModel.load(MODEL_PATH)


MODEL = ensure_model()


def predict_price(
    overall_qual: int,
    gr_liv_area: int,
    garage_cars: int,
    total_bsmt_sf: int,
    full_bath: int,
    year_built: int,
    neighborhood: str,
    house_style: str,
) -> str:
    prediction = MODEL.predict(
        {
            "OverallQual": overall_qual,
            "GrLivArea": gr_liv_area,
            "GarageCars": garage_cars,
            "TotalBsmtSF": total_bsmt_sf,
            "FullBath": full_bath,
            "YearBuilt": year_built,
            "Neighborhood": neighborhood,
            "HouseStyle": house_style,
        }
    )
    return f"${prediction:,.0f}"


with gr.Blocks(title="House Price Predictor") as demo:
    gr.Markdown("# House Price Predictor")
    gr.Markdown("Predict sale prices using a model trained on the Kaggle House Prices dataset.")

    with gr.Row():
        with gr.Column():
            overall_qual = gr.Slider(1, 10, value=7, step=1, label="Overall quality")
            gr_liv_area = gr.Number(value=1800, label="Above-ground living area")
            garage_cars = gr.Slider(0, 5, value=2, step=1, label="Garage capacity")
            total_bsmt_sf = gr.Number(value=1000, label="Total basement square feet")
        with gr.Column():
            full_bath = gr.Slider(0, 5, value=2, step=1, label="Full bathrooms")
            year_built = gr.Number(value=1995, label="Year built")
            neighborhood = gr.Dropdown(
                ["NAmes", "CollgCr", "OldTown", "Edwards", "Somerst", "NridgHt", "Gilbert", "NoRidge"],
                value="Somerst",
                label="Neighborhood",
                allow_custom_value=True,
            )
            house_style = gr.Dropdown(
                ["1Story", "2Story", "1.5Fin", "SLvl", "SFoyer"],
                value="2Story",
                label="House style",
                allow_custom_value=True,
            )

    output = gr.Textbox(label="Predicted sale price", interactive=False)
    predict_button = gr.Button("Predict", variant="primary")
    predict_button.click(
        fn=predict_price,
        inputs=[
            overall_qual,
            gr_liv_area,
            garage_cars,
            total_bsmt_sf,
            full_bath,
            year_built,
            neighborhood,
            house_style,
        ],
        outputs=output,
        api_name="predict",
    )

    gr.Markdown(
        f"Model metrics: RMSE `{MODEL.metrics.get('rmse')}`, "
        f"MAE `{MODEL.metrics.get('mae')}`, R2 `{MODEL.metrics.get('r2')}`"
    )


if __name__ == "__main__":
    demo.launch()