# Mixed Effect Modeling and Variable Selection for Quantile Regression
# Bar, Booth, and Wells (Statistical Modelling 2021)

The zip file which includes this description contains code to run
simulation scenarios from the paper.

The QREM package is available from github:
https://github.com/haimbar/QREM

The QREM package uses SEMMS for variable selection when the number of
predictors is large. SEMMS is available from
# https://github.com/haimbar/SEMMS

In order to initialize the set of selected predictors, SEMMS uses the
edgefinder package:
# https://github.com/haimbar/edgefinder
edgefinder detects highly correlated pairs of predictors. It is used
by SEMMS to select a subset of the variables so as to prevent multi
collinearity. When P is large, the calculations made by edgefinder
may take a long time, and may require a large amount of memory. 
In such cases, the matrix of predictors may be broken into several
matrices (e.g., with 1000 columns each) and the QR model can be estimated
separately with each matrix, in order to obtain the important predictors.

The files Fig2.R-Fig4.R were used in the low-dimensional simulation
scenarios. The files Table1.R and Table2.R were used in the variable
selection scenarios.

Some simulations, especially ones involving the bootstrap may take a
long time.

Visit the github repositories in order to check for software/dat updates.

