import gradio as gr
import pandas as pd
import joblib
import numpy as np

# Load the trained model
model = joblib.load('obesity_prediction_model.pkl')

# Define the obesity level descriptions
OBESITY_DESCRIPTIONS = {
    'Insufficient_Weight': 'Underweight (BMI < 18.5)',
    'Normal_Weight': 'Normal Weight (BMI 18.5 - 24.9)',
    'Overweight_Level_I': 'Overweight Level I (BMI 25.0 - 27.4)',
    'Overweight_Level_II': 'Overweight Level II (BMI 27.5 - 29.9)',
    'Obesity_Type_I': 'Obesity Type I (BMI 30.0 - 34.9)',
    'Obesity_Type_II': 'Obesity Type II (BMI 35.0 - 39.9)',
    'Obesity_Type_III': 'Obesity Type III (BMI >= 40)'
}


def predict_obesity(
    gender: str,
    age: float,
    family_history: str,
    favc: str,
    fcvc: float,
    ncp: float,
    caec: str,
    smoke: str,
    ch2o: float,
    scc: str,
    faf: float,
    tue: float,
    calc: str,
    mtrans: str
) -> str:
    """
    Predict obesity level based on input features.

    Returns:
        str: Predicted obesity level with description
    """
    # Create input dataframe
    input_data = pd.DataFrame({
        'Gender': [gender],
        'Age': [age],
        'family_history_with_overweight': [family_history],
        'FAVC': [favc],
        'FCVC': [fcvc],
        'NCP': [ncp],
        'CAEC': [caec],
        'SMOKE': [smoke],
        'CH2O': [ch2o],
        'SCC': [scc],
        'FAF': [faf],
        'TUE': [tue],
        'CALC': [calc],
        'MTRANS': [mtrans]
    })

    # Make prediction
    prediction = model.predict(input_data)[0]
    probabilities = model.predict_proba(input_data)[0]
    confidence = np.max(probabilities) * 100

    # Get description
    description = OBESITY_DESCRIPTIONS.get(prediction, prediction)

    # Format result
    result = f"**Predicted Obesity Level:** {prediction}\n\n"
    result += f"**Description:** {description}\n\n"
    result += f"**Confidence:** {confidence:.1f}%"

    return result


# Create Gradio interface
with gr.Blocks(title="Obesity Level Prediction", theme=gr.themes.Soft()) as demo:
    gr.Markdown(
        """
        # Obesity Level Prediction
        ### Based on Eating Habits and Lifestyle Factors

        This application predicts obesity levels using machine learning based on your eating habits,
        physical activity, and lifestyle information.

        ---
        """
    )

    with gr.Row():
        with gr.Column():
            gender = gr.Dropdown(
                choices=["Female", "Male"],
                label="Gender",
                value="Female"
            )
            age = gr.Number(
                label="Age",
                value=25,
                minimum=10,
                maximum=100
            )
            family_history = gr.Dropdown(
                choices=["yes", "no"],
                label="Family history with overweight",
                value="yes"
            )
            favc = gr.Dropdown(
                choices=["yes", "no"],
                label="Frequent consumption of high caloric food (FAVC)",
                value="no"
            )
            fcvc = gr.Slider(
                minimum=1,
                maximum=3,
                step=0.1,
                label="Frequency of consumption of vegetables (FCVC) [1-3]",
                value=2
            )
            ncp = gr.Slider(
                minimum=1,
                maximum=4,
                step=0.1,
                label="Number of main meals (NCP) [1-4]",
                value=3
            )
            caec = gr.Dropdown(
                choices=["no", "Sometimes", "Frequently", "Always"],
                label="Consumption of food between meals (CAEC)",
                value="Sometimes"
            )

        with gr.Column():
            smoke = gr.Dropdown(
                choices=["yes", "no"],
                label="Smoking habit (SMOKE)",
                value="no"
            )
            ch2o = gr.Slider(
                minimum=1,
                maximum=3,
                step=0.1,
                label="Consumption of water daily (CH2O) [1-3 liters]",
                value=2
            )
            scc = gr.Dropdown(
                choices=["yes", "no"],
                label="Calories consumption monitoring (SCC)",
                value="no"
            )
            faf = gr.Slider(
                minimum=0,
                maximum=3,
                step=0.1,
                label="Physical activity frequency (FAF) [0-3 days/week]",
                value=1
            )
            tue = gr.Slider(
                minimum=0,
                maximum=2,
                step=0.1,
                label="Time using technology devices (TUE) [0-2 hours]",
                value=1
            )
            calc = gr.Dropdown(
                choices=["no", "Sometimes", "Frequently", "Always"],
                label="Consumption of alcohol (CALC)",
                value="Sometimes"
            )
            mtrans = gr.Dropdown(
                choices=["Automobile", "Motorbike", "Bike", "Public_Transportation", "Walking"],
                label="Transportation used (MTRANS)",
                value="Public_Transportation"
            )

    predict_btn = gr.Button("Predict Obesity Level", variant="primary")

    output = gr.Markdown(label="Prediction Result")

    predict_btn.click(
        fn=predict_obesity,
        inputs=[
            gender, age, family_history, favc, fcvc, ncp, caec,
            smoke, ch2o, scc, faf, tue, calc, mtrans
        ],
        outputs=output
    )

    gr.Markdown(
        """
        ---
        ### Obesity Level Categories:
        - **Insufficient Weight:** BMI < 18.5
        - **Normal Weight:** BMI 18.5 - 24.9
        - **Overweight Level I:** BMI 25.0 - 27.4
        - **Overweight Level II:** BMI 27.5 - 29.9
        - **Obesity Type I:** BMI 30.0 - 34.9
        - **Obesity Type II:** BMI 35.0 - 39.9
        - **Obesity Type III:** BMI >= 40

        ---
        *CSC14119 - Introduction to Data Science - DIY 2*
        """
    )

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