Coordinated descent in linear models project discussion, summer 2012

1. refactoring of ElasticNet and Lasso classes
2. setting up tests to benchmark current and future code
3. cd speed up#
   1. important equations
   2. efficient implementation of covariance updates
   3. l2 loss
   4. log loss
   5. multi-logit loss

Midterm Evaluation (July 13)

1. strong rule implementation
   1. sequential strong rule
2. extending LARS
3. parallelization ( matrix of outcome )
4. narrative documentation

Final Evaluation (Aug 24)