This week was spent in further research into testing methods for the implementations in the PR(#1037). Several runs and experiments are in the Gists linked below :
- Sparse Matrix completion synthetic run #1
- Sparse Matrix completion synthetic run #2
- Sparse Matrix completion run on sampled data
- Sparse SVM run on RCV1 dataset
Also, I have started with the initial implementation of another optimizer method, Stochastic Co Ordinate descent, and would open up a PR soon with the additions.