Automatic model training and testing in GitHub action

(Continuous integration in Machine Learning)

Why we need CI in machine learning?

Machine learning is a very complicate workflow, including data processing, data merging, data modeling, data evaluation, etc.

It takes months to build a useful model, so how can we decrease the time of build model to make different experiment. Automating the process of machine learning is a nice way. CI in machine learning probably can help this, CI in ML make our model more automatic and more easily to check.

This process will make each elements of complicated machine learning model more trackable, including data, code, model result, model structure.

What is the essential steps in CI in machine learning in GitHub action?

1. make Yaml file in Pull request, yaml file will control you want to print on the report.

2. make Python scripts about the training model.

3. pull request the yaml file and see the results on GitHub pull request.

reference : https://www.youtube.com/watch?v=9BgIDqAzfuA