Confidence sets of models

Software implementing the procedure of Cox and Battey (2017), Proc. Nat. Acad. Sci., 114 (32), 8592-8595 is available in R and Matlab.


Matlab code for reproducing the analysis of Cox and Battey (2017), Proc. Nat. Acad. Sci., 114 (32), 8592-8595.

OA.m Runs everything. Calls glmRoutine.m
OAData.mat Data on 129 females, 105 cases, 24 controls.
varNames.csv Names of the variables in OAData.mat



Matlab code for reproducing the simulation of Battey and Cox (2018), Proc. R. Soc. A, 474, 20170631.

simulations.m Reproduces the data in Table 1. Calls the following functions.
phase1OnlySelect2.m First stage reduction. Calls regress_hsb.m.
phase2Only.m Second stage reduction.
fTestModelsv2.m Constructs `confidence sets' of models.
combntns.m Finds all low dimensional subsets of models for test.













Matlab code for reproducing the simulation of Lewis and Battey (2023).

The analysis of Lewis and Battey (2023) suggested improvements to the procedure beyond those indicated by Battey and Cox (2018). These are available from R. M. Lewis's github page.



R software

The procedure has been implemented in the R package `HCmodelSets', translated from the original matlab source by H. H. Hoeltgebaum and available on the Comprehensive R Archive Network. A detailed guide to usage is:

Hoeltgebaum, H. H. and Battey, H. S. (2019)
HCmodelSets: An R package for specifying sets of well-fitting models in high dimensions.
The R Journal, 11, 370-379.