The goal of MarginalMaxTest is to test the marginal correlation between a scalar response variable with a vector of explanatory variables using the max-type test with bootstrap. The test is based on the max-type statistic and its asymptotic distribution under the null hypothesis of no marginal correlation. The bootstrap procedure is used to approximate the null distribution of the test statistic. The package provides a function for performing the test. For more technical details, refer to Zhang 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 from GitHub 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 data
set.seed(47)
n <- 200
p <- 10
x <- matrix(rnorm(n*p), n, p)
y <- 0.25*x[,1] + rnorm(n)
# Run the test
marginal.test(x, y, B = 200, method = "adaptive")
#> $p_value
#> [1] 0.01
#>
#> $time
#> [1] 0.007
marginal.test(x, y, B = 200, method = "max")
#> $p_value
#> [1] 0
#>
#> $time
#> [1] 0.007
marginal.test(x, y, B = 200, method = "sum")
#> $p_value
#> [1] 0.11
#>
#> $time
#> [1] 0.007