Branches for a data science project

Question

I am working on a data science project in my git repository.

It mainly contains 2 things:

1.  A package that I am developing
2. Experiments, which include scripts that use the packages and tell something about the data. 

 I have completed doing 2 branches which are:

1. Development - It contains new releases of the package which are prepared on the branch. for example, new features to be added or API to be changed, etc.
2. Master-It included the current and working release of the package. So that everyone can download the package and work on it.

I am experimenting on the master branch, I found it a bit weird. Can anyone guide me with an alternative? Should I perform a branch experiment?

Answer 1

As per my knowledge you are asking about the workflow and branch management strategy.

Git workflows are very common and have very little value. In case if you have a small team then you can use trunk-based development. where you and other developers can share the branch under source-control.

If in case you have any plan of quality control mechanism on the experiment then you probably have to keep the experiment out of the master, until it is verified. You need to tests the branching strategy and run via CI prior for approving pull request

If you want to know more about data science then do check out the following data science blog which will help you in understanding data science from scratch