🩺 Diabetes Risk Prediction App

Overview

This project is a Web Application that predicts the probability of diabetes based on diagnostic measures.

It is based on a Comparative Analysis of Naive Bayes classifiers, where Gaussian Naive Bayes was identified as the most accurate model (90.48% Accuracy) for this dataset.

🔗 Live Demo

➡️ Click Here to Open the App

Methodology

We compared three variants of Naive Bayes:

1. Gaussian NB: Best for continuous features (Glucose, BMI). (Selected Model)
2. Bernoulli NB: Best for binary features.
3. Multinomial NB: Best for count data.

Key Metrics

Model	Accuracy
Gaussian NB	90.48%
Bernoulli NB	~88%
Multinomial NB	~76%

Installation & Usage

1. Clone the repository:

   git clone [https://github.com/Muhammad-Shahan/Diabetes-Risk-Prediction.git](https://github.com/Muhammad-Shahan/Diabetes-Risk-Prediction.git)

2. Install dependencies:

   pip install -r requirements.txt

3. Run the App:

   streamlit run app.py

Project Structure

• app.py: The main Streamlit interface.
• train_model.py: Script used to train and save the model.
• analysis.ipynb: Jupyter Notebook containing the research and EDA.
• diabetes_model.pkl: The trained GaussianNB model file.
• diabetes_prediction_dataset.csv: The dataset used for training.

Citation

If you find this useful, please star the repo!