Description Usage Arguments Details References Examples
Hierarchical inference testing for linear models with highdimensional and/or correlated covariates by repeated sample splitting.
1 2 3 4 
x 
Design matrix of dimension 
y 
Quantitative response variable dimension 
hierarchy 
Object of class 
family 
Family of response variable distribution. Ether 
B 
Number of samplesplits. 
p.samp1 
Fraction of data used for the LASSO. The hierachical ANOVA testing uses

nfolds 
Number of folds (default is 10). See 
overall.lambda 
Logical, if true, lambda is estimated once, if false, lambda is estimated for each sample split. 
lambda.opt 
Criterion for optimum selection of crossvalidated lasso. Either
"lambda.1se" (default) or "lambda.min". See 
alpha 
A single value in the range of 0 to 1 for the elastic net mixing parameter. 
gamma 
Vector of gammavalues. 
max.p.esti 
Maximum alpha level. All pvalues above this value are set to one.
Small 
mc.cores 
Number of cores for parallelising. Theoretical maximum is 'B'. For
details see 
trace 
If TRUE it prints current status of the program. 
... 
Additional arguments for 
The H0model contains variables, with are not tested, like experimentaldesign variables. These variables are not penalised in the LASSO model selection and are always include in the reduced ANOVA model.
Mandozzi, J. and Buehlmann, P. (2013). Hierarchical testing in the highdimensional setting with correlated variables. To appear in the Journal of the American Statistical Association. Preprint arXiv:1312.5556
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18  # Simulation:
set.seed(123)
n < 80
p < 82
## x with correlated columns
corMat < toeplitz((p:1/p)^2)
corMatQ < chol(corMat)
x < matrix(rnorm(n * p), nrow = n) %*% corMatQ
colnames(x) < paste0("x", 1:p)
## y
mu < x[, c(5, 10, 72)] %*% c(2, 2, 2)
y < rnorm(n, mu)
## clustering of the clumns of x
hc < hclust(dist(t(x)))
# HIT with AF
out < hit(x, y, hc)
summary(out)

Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.