Repository files navigation

The goal of MarginalMaxTest is to test the marginal correlation betweena scalar response variable with a vector of explanatory variables usingthe max-type test with bootstrap. The test is based on the max-typestatistic and its asymptotic distribution under the null hypothesis ofno marginal correlation. The bootstrap procedure is used to approximatethe null distribution of the test statistic. The package provides afunction for performing the test. For more technical details, refer toZhang and Laber (2014) doi:10.1080/01621459.2015.1106403.

Install the R CRAN version of MarginalMaxTest like so:

install.packages("MarginalMaxTest")

You can also install the development version of MarginalMaxTest fromGitHub with:

# install.packages("devtools")devtools::install_github("canyi-chen/MarginalMaxTest")

This is a basic example which shows you how to solve a common problem:

library(MarginalMaxTest)# Generate sample dataset.seed(47)n<-200p<-10x<-matrix(rnorm(n*p),n,p)y<-0.25*x[,1]+ rnorm(n)# Run the testmarginal.test(x,y,B=200,method="adaptive")#> $p_value#> [1] 0.01#>#> $time#> [1] 0.007marginal.test(x,y,B=200,method="max")#> $p_value#> [1] 0#>#> $time#> [1] 0.007marginal.test(x,y,B=200,method="sum")#> $p_value#> [1] 0.11#>#> $time#> [1] 0.007