Example of ML model implementation in VoltDB to predict delays in flights. The library H2O is used to build the model since it's a market leader in machine learning and also provides the convenience of exporting models as Java code
Steps to run (assuming you have VoltDB and H2O installed)
- Create a new flow or load a flow from h2o/ to start off a previously built model.
- Train the model using your own data or data from data/ (using the H2O Flow UI or a language SDK)
- Export the trained model as Java class.
- Create a new stored procedure that accepts new events and calls the trained model class to make the predictions
In this project, I have implemented a Linear Regression Model for Flight Prediction. On the test set, this model only gets about 69% accuracy but it's a place to start.