app.py

import gradio as gr
import pandas as pd
import xgboost as xgb
from huggingface_hub import hf_hub_download

# Download and load the model from the Hugging Face repository
model_path = hf_hub_download(
    repo_id='caslabs/xgboost-home-price-predictor',
    filename='xgboost_model.json',
    repo_type='model'
)

loaded_model = xgb.XGBRegressor()
loaded_model.load_model(model_path)

# Define the prediction function
def predict_price(data):
    # Parse the input data
    input_data = pd.DataFrame([data])

    # Make a prediction
    predicted_price = loaded_model.predict(input_data)[0]
    return {"predicted_price": round(predicted_price, 2)}

# Define the API interface with Gradio
iface = gr.Interface(
    fn=predict_price,
    inputs=gr.JSON(),  # Accept JSON input
    outputs=gr.JSON(),  # Return JSON output
    title="Home Price Prediction API",
    description="Provide property details in JSON format, and the model will return the predicted adjusted sale price."
)

iface.launch()