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

Robustness weights in sparseS() #29

Closed aalfons closed 3 years ago

aalfons commented 9 years ago

Current implementation of weights in FAST-S algorithm is only valid for lambda different from 0.

For a fixed lambda grid, we need to check for special case lambda = 0 and compute weights as for the unpenalized S-estimator in that case.

When lambda is selected on each weighted subset, we have to exclude lambda = 0 in order to select lambda only from values where the weights are appropriate.

voellerer commented 9 years ago

Implementation should also be correct for lambda equal to zero. See derivations in email.

aalfons commented 3 years ago

sparseS() will no longer be developed further. Users can use package pense instead.