After diagnosed with diabetes at 2020, I was curious to create a small project for diabetes early detection. To achieve that of course I need machine learning and the dataset of diabetes symptoms and the diagnosis results. Fortunately the dataset is available at https://archive.ics.uci.edu/ml/datasets/Early+stage+diabetes+risk+prediction+dataset. What I need next is to build the prediction system and my choice for that of course Python with scikit-learn library.

Personally, what makes this project interesting is rather than just creating the machine learning model, this application is the real implementation with four user interfaces: interactive text mode (CLI), web browser based (Web), graphical/desktop based (GUI), and also API mode (RESTAPI). From those four interfaces, user must input the age, gender, then choose some known diabetes symptoms. After submitting the symptoms, the application will predict the result if it's POSITIVE or NEGATIVE.

The algorithm used for the machine learning is SVM (Support Vector Machine). Before generating the machine learning model, the application used grid search cross validation for determining the most optimal parameter for the SVM algorithm. The trained model can also be saved into a file so the application can load it later and doesn't have to repeat the training process.

More detailed technical information about this project and its source code for this project is available at the following GitHub repository: https://github.com/w3shaman/diabetes-detection.

For easier use and distribution, I also made the docker image for it. Please check it out at: https://hub.docker.com/r/w3shaman/diabetes-detection.

Disclaimer
This application is for learning and experiment purpose only. The prediction result may not meet your expectation.