app.py

import gradio as gr
import pandas as pd
import joblib
import os

# Load the trained Naïve Bayes model
model_path = os.path.join(os.getcwd(), "Random Forest_model.pkl")
model = joblib.load(model_path)

# Define the prediction function
def predict_status(age, gender, family_income, attendance, grades, paying_school_feeding, feeding_rate, orphans, parental_education, number_of_siblings):
    try:
        # Create a dictionary from inputs
        input_data = {
            'age': age,
            'gender': gender,
            'family_income': family_income,
            'attendance': attendance,
            'grades': grades,
            'paying_school_feeding': paying_school_feeding,
            'feeding_rate': feeding_rate,
            'orphans': orphans,
            'parental_education': parental_education,
            'number_of_siblings': number_of_siblings
        }

        # Convert input to DataFrame
        new_data = pd.DataFrame([input_data])

        # Encoding categorical variables
        encoding_maps = {
            'gender': {'Male': 0, 'Female': 1},
            'paying_school_feeding': {'Yes': 1, 'No': 0},
            'orphans': {'Yes': 1, 'No': 0},
            'parental_education': {'Not_Ediucated': 0, 'Primary': 1, 'Secondary': 2, 'University': 3},
        }

        for col, mapping in encoding_maps.items():
            if col in new_data:
                new_data[col] = new_data[col].map(mapping)

        # Ensure all values are numeric
        if new_data.isnull().values.any():
            return "Error: Invalid input values. Please check your inputs."

        # Make a prediction
        prediction = model.predict(new_data)

        # Decode prediction result
        status_map = {0: 'Studying', 1: 'Dropout'}
        predicted_status = status_map.get(prediction[0], "Unknown")

        return f"Predicted Status: {predicted_status}"

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

# Define Gradio interface
inputs = [
    gr.Number(label="Age"),
    gr.Dropdown(label="Gender", choices=["Male", "Female"]),
    gr.Number(label="Family Income"),
    gr.Number(label="Attendance (0-100)"),
    gr.Number(label="Grades (0-100)"),
    gr.Dropdown(label="Paying School Feeding", choices=["Yes", "No"]),
    gr.Number(label="Feeding Rate (0-4)"),
    gr.Dropdown(label="Orphans", choices=["Yes", "No"]),
    gr.Dropdown(label="Parental Education", choices=["Not_Ediucated", "Primary", "Secondary", "University"]),
    gr.Number(label="Number of Siblings")
]

outputs = gr.Textbox(label="Prediction Result")

# Create Gradio app
app = gr.Interface(
    fn=predict_status,
    inputs=inputs,
    outputs=outputs,
    title="Student Dropout Prediction",
    description="Predict whether a student will drop out or continue studying based on various features."
)

# Launch the app
app.launch()