diabetes_project.py

# -*- coding: utf-8 -*-
"""app.py

Automatically generated by Colab.

Original file is located at
    https://colab.research.google.com/drive/1ryMEIbhOIiyQDhLvzCXtBpTmBQ3vWrTq
"""

import gradio as gr
import pickle
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# Load the trained model
with open("model3.pkl", "rb") as f:
    model = pickle.load(f)

# Prediction function
def predict_diabetes(pregnancies, glucose, bp, skin, insulin, bmi, dpf, age):
    input_data = np.array([[pregnancies, glucose, bp, skin, insulin, bmi, dpf, age]])
    prediction = model.predict(input_data)[0]

    # Create Seaborn plot
    plt.clf()
    plt.figure(figsize=(4, 3))
    sns.barplot(x=["Diabetes Risk Prediction"], y=[prediction], palette="mako")
    plt.title("Model Output")
    plt.ylabel("Prediction (0 = No, 1 = Yes)")
    plot_path = "prediction_plot.png"
    plt.savefig(plot_path)
    plt.close()

    return int(prediction), plot_path

# Gradio Interface
gr.Interface(
    fn=predict_diabetes,
    inputs=[
        gr.Slider(0, 20, step=1, label="Pregnancies"),
        gr.Slider(0, 200, label="Glucose"),
        gr.Slider(0, 150, label="Blood Pressure"),
        gr.Slider(0, 100, label="Skin Thickness"),
        gr.Slider(0, 900, label="Insulin"),
        gr.Slider(10.0, 70.0, step=0.1, label="BMI"),
        gr.Slider(0.0, 2.5, step=0.01, label="Diabetes Pedigree Function"),
        gr.Slider(10, 100, label="Age")
    ],
    outputs=[
        gr.Textbox(label="Prediction (0 = No Diabetes, 1 = Diabetes)"),
        gr.Image(type="filepath", label="Prediction Plot")
    ],
    title="Diabetes Prediction Based on Health Data"
).launch()