Dimensionality reduction using PCA

[bkowshik]

We currently have 115 attributes for every changeset. It is practically impossible how all these features look like. But, we could use techniques of Dimensionality reduction and project them into 2 dimensions.

We have powerful Principal component analysis (PCA) as part of scikit-learn:

• http://scikit-learn.org/stable/modules/decomposition.html