Resampling Penalization for histogram selection in regression

Last updated: October, 31st 2008 - Back to index

Description

Resampling Penalization is a family of model selection procedure by penalization that can use any exchangeable weighted bootstrap resampling scheme to compute a penalty. It is properly defined in the general framework and extensively studied for histogram selection in regression in the following paper:
(2009) Sylvain Arlot. Model selection by resampling penalization. Electronic Journal of Statistics, 3, (2009), 557-624 (electronic). DOI: 10.1214/08-EJS196.
[journal]   [appendix, pdf]   [Hal]   [slides EMS 2009]

The present software allows to perform this algorithm for several examples of weights in the histogram selection case.
This code is not meant to be fast, but mainly to allow reproducibility of the results presented in the above paper.

Matlab code - version 1.0

The code requires the Statistics toolbox. The archive contains three matlab files. RPmodsel_hist.m is a function that performs model selection by resampling penalization among a collection of histogram models. All the examples of weights given in the paper can be used, and the penalties can either be computed from exact formulas given in the paper or by Monte-Carlo approximation.
A demo of how to use the function RPmodsel_hist and simulate data with gener_data.m is provided with the file main.m.

Resampling penalization - version 1.0


The Resampling Penalization package is provided free for non-commercial use under the terms of the GNU General Public License. If you have any questions or comments regarding this package, or if you want to report any bugs, please send an e-mail to sylvain.arlotREMOVETHIS@u-psud.fr. The current version 1.0 has been released on October 31st 2008. Check this web page regularly for newer versions.

THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.