|
Least Angle Regression software: LARS
"Least Angle Regression" ("LAR") is a new model
selection
algorithm; a useful and less greedy version of
traditional
forward selection methods. LAR is described in detail in a paper
by Brad Efron, Trevor Hastie, Iain Johnstone and Rob Tibshirani,
soon to
appear in the Annals of Statistics.
A simple modification of the LAR algorithm
implements Tibshirani's
Lasso, an attractive version of OLS that constrains
the sum of the
absolute regression coefficients; the Lasso modification of
the LARS
software calculates the entire Lasso path of coefficients for a
given
problem at the cost of a single least squares fit.
A different LARS modification efficiently
implements epsilon Forward
Stagewise linear regression, another promising new
model selection
method closely related to Boosting.
--------------------------------------------------------------------
Trevor
Hastie
hastie@stanford.edu
Professor, Department of Statistics, Stanford University
Phone: (650)
725-2231 (Statistics) Fax:
(650) 725-8977
(650) 498-5233 (Biostatistics) Fax:
(650) 725-6951
URL: http://www-stat.stanford.edu/~hastie
address: room 104, Department of Statistics, Sequoia
Hall
390 Serra
Mall, Stanford University, CA 94305-4065
--------------------------------------------------------------------
|