An end-to-end machine learning application that predicts whether a patient has diabetes based on diagnostic measurements. This project includes data analysis, model development, and deployment on Heroku as a web application.
Diabetes is a chronic health condition affecting millions of people worldwide. Early detection is crucial for effective management. This project utilizes machine learning to:
- Analyze diagnostic measurements from patients
- Identify patterns associated with diabetes
- Provide quick risk assessment without invasive tests
- Generate downloadable prediction reports for healthcare practitioners
The model is trained on the Pima Indians Diabetes Database, which includes diagnostic measurements for:
- Number of pregnancies
- Glucose concentration
- Blood pressure
- Skin thickness
- Insulin level
- BMI (Body Mass Index)
- Diabetes pedigree function (diabetes hereditary factor)
- Age
Each record is classified as having diabetes (1) or not having diabetes (0).
- Handling missing values
- Detecting and managing outliers
- Feature scaling and normalization
- In-depth exploratory data analysis (EDA)
The project implements multiple algorithms for comparison:
- Logistic Regression
- Random Forest
- Support Vector Machine
- K-Nearest Neighbors
- Gradient Boosting
After evaluation, the best performing model is selected for deployment.
Performance metrics include:
- Accuracy
- Precision
- Recall
- F1 Score
- ROC-AUC Curve
The Flask-based web application features:
- User-friendly interface for entering diagnostic measurements
- Real-time prediction feedback
- Downloadable prediction results in CSV format
- Comprehensive EDA visualization dashboard
The application is deployed on Heroku using:
- Git integration for seamless deployment
- Flask server configuration
- Gunicorn WSGI server
- Environment variables for secure configuration
The exploratory data analysis revealed:
- Strong correlation between glucose levels and diabetes diagnosis
- BMI as a significant predictor for diabetes risk
- Age-related patterns in diabetes occurrence
- Important hereditary factors reflected in the diabetes pedigree function
- Language: Python 3.7
- Machine Learning: scikit-learn, pandas, numpy
- Data Visualization: matplotlib, seaborn
- Web Framework: Flask
- Deployment: Heroku
- Version Control: Git, GitHub
- IDE: Jupyter Notebook
# Clone the repository
git clone https://github.com/Dishant27/Diabetes_Predictor.git
cd Diabetes_Predictor
# Set up a virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Run the application
python app.pyAfter running, the application will be available at http://localhost:5000.
- Implement more advanced models like Neural Networks
- Add feature importance visualization
- Incorporate confidence intervals for predictions
- Develop a mobile application version
- Integrate with additional health metrics
- Add user authentication for healthcare providers
- Implement longitudinal tracking for patient monitoring over time
- Pima Indians Diabetes Database
- Flask Documentation
- scikit-learn Documentation
- Heroku Deployment Guide
Dishant - GitHub Profile
Note: This project is for educational purposes and should not replace professional medical advice. Always consult healthcare providers for diabetes diagnosis and treatment.
Enhanced Documentation