app.py

from io import StringIO
import gradio as gr
import pandas as pd
import xgboost as xgb
  # Add this import

def preprocess_data(df):


    # Assuming the model expects the input without the first 50 columns
    processed_data = df.drop(df.columns[0:50], axis=1)

    # Assuming you want the last column as the target.
    X = processed_data.drop(processed_data.columns[-1], axis=1).values
    y = processed_data[processed_data.columns[-1]].values


    return (X, y)



def process_csv_text(temp_file):
    if isinstance(temp_file, str):
      df = pd.read_csv(StringIO(temp_file))
    else:
      df = pd.read_csv(temp_file.name)
    print(df)
    return df


def predict_interface(input_csv):
    print(type(input_csv))
    loaded_model = xgb.XGBRegressor()
    loaded_model.load_model('xgboost_model.json')

    # Check if the file is non-empty and has recognizable data
    try:
        # Load the csv content into a DataFrame
        data = process_csv_text(input_csv)
        if data is None:
            return "Uploaded file is empty or not recognized as a valid CSV."

        # Continue with the rest of the processing
        processed_data = preprocess_data(data)
        predictions = loaded_model.predict(processed_data[0])  # Access the X part
        data['Predictions'] = predictions
        data['Actual'] = processed_data[1]  # Access the y part
        return data.to_html()
    except Exception as e:
        return str(e)


# Define the Gradio interface
iface = gr.Interface(
    fn=predict_interface,
    inputs="file",
    outputs="html",
    live=True
)

iface.launch(server_name="0.0.0.0", server_port=7860)