search

aalfons / robustHD

Robust methods for high-dimensional data, in particular linear model selection techniques based on least angle regression and sparse regression.
GNU General Public License v3.0
10 stars 6 forks source link

Cross-validation from within model fitting functions #7

Closed aalfons closed 12 years ago

aalfons commented 12 years ago

Allow for cross-validation from within model fitting functions via the crit argument.

aalfons commented 12 years ago

Full object from cvTools should be stored to take advantage of the more advanced plot methods. critPlot() should in this case just be a wrapper.

aalfons commented 12 years ago

sparseLTSGrid() and fitModels() should be adjusted such that only the final model is computed from the full data set to save computation time. Class structure and methods then probably need to be adapted, too.

aalfons commented 12 years ago

It's probably a good idea to switch to perry directly and allow for random splits and bootstrap prediction error estimation as well.

aalfons commented 12 years ago

Class structure should be modified such that ocefficients etc. in this case are only stored for the optimal model. Hence the coef() method etc. should always ignore the argument s or set it to NULL.

aalfons commented 12 years ago

It's a bit trickier for sparseLTSGrid() because the optimal model for the reweighted and raw estimator can be different.