Pima Indians Diabetes Classification

This project focuses on developing a predictive model to diagnose diabetes in Pima Indians using machine learning techniques. The dataset includes various medical predictor variables and a target variable indicating diabetes presence.

Dataset

• Features: Includes medical metrics such as the number of pregnancies, BMI, insulin levels, age, and more.
• Target: Diabetes outcome (0 = No, 1 = Yes).

Process

1. Exploratory Data Analysis (EDA)

   • Conducted comprehensive data cleansing.
   • Performed detailed analysis of each feature to ensure data quality and integrity.

2. Feature Engineering

   • Created new features and assessed their impact using SHAP (Shapley Additive Explanations).

3. Synthetic Minority Oversampling Technique (SMOTE)

   • Applied SMOTE to address class imbalance by generating synthetic samples for the minority class, enhancing model performance.

4. Model Implementation

   • Random Forest: Achieved 93% accuracy. Feature importance was analyzed using SHAP, revealing that low insulin levels and the interaction between age and insulin are highly influential.
   • Deep Learning Model: Evaluated for performance; results were comparable to the Random Forest model.

5. Deployment

   • Model Selection: Random Forest was chosen for deployment due to its interpretability and efficiency.
   • Streamlit Application: Developed for easy prediction, with integrated preprocessing and feature engineering.

Key Insights

• Influential Features: Low insulin values and the interaction between age and insulin significantly affect predictions.
• Less Impactful Features: Features like blood pressure and number of pregnancies have minimal effect and can be omitted to reduce computational costs.

Installation

To install the necessary dependencies, run:

pip install -r requirements.txt

Usage

1. Run the Streamlit Application

   streamlit run app.py

2. Input Data

   • Prepare a pandas DataFrame with the same structure as the training dataset.
   • The Streamlit app will handle preprocessing and feature engineering, then provide predictions.

Contributing

Contributions are welcome! If you have suggestions, bug reports, or feature requests, please open an issue or submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.