How do I migrate a flat svn repo to git repo?

How does a data scientist use version control tools like GIT and SVN?

• Usually data science project involves in using a lot of of-the-shelf tools and packages in their source code. The amount of code written by them is usually not high as compared to that of a typical software development project(say web application development). Assuming a data scientist involved in development data preparation, model development and deployment, what should be the motivating factor to learn a tool like git or svn? Some real examples in day to day work will also help. Of course he doesn't want to loose his code and periodically backs them up. If you are asked to educate a group of them, what are the first few points that you would like to tell them?

• Answer:

  Personally I don't use Git for everyday's small small tasks. But once I am working for a project and I have formalized the possible course of action then I create a Git project and commit as the project evolves. This gives me control over the project and confidence that I am moving forward. There are times when multiple people are working on same project code. Recently I was working on a data science product for my company and there were 3 people working on the same project. I was writing all the Machine Learning/Data Mining code, other team members were building the integration code which takes my code and fit it with other systems.  So, everyone was suppose to have the same code which they checkout make their own branch and then commit their code time to time for others. Now more people can join the development easily. I think the concept of version control is good but should not be overused in Data Science for every small changes. Keep it for only committing project milestones.