Skip to content

Coordinated descent in linear models project discussion, summer 2012

ibayer edited this page Jul 2, 2012 · 12 revisions
Clone this wiki locally
  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)

Something went wrong with that request. Please try again.