Test association between sparse leading eigen-value and variable

Score impact of each sample on sparse leading eigen-value and then peform test of association with variable using non-parametric test

Usage

sle.test(
  Y,
  variable,
  method = c("pearson", "kendall", "spearman"),
  rho = 0,
  sumabs = 1
)

Arguments

Y

data matrix with samples on rows and variables on columns

variable

variable with number of entries must equal nrow(Y). Can be discrete or continuous.

method

specify which correlation method: "pearson", "kendall" or "spearman"

rho

a positive constant such that cor(Y) + diag(rep(rho,p)) is positive definite.

sumabs

regularization paramter. Value of 1 gives no regularization, sumabs*sqrt(p) is the upperbound of the L_1 norm of v,controling the sparsity of solution. Must be between 1/sqrt(p) and 1.

Value

list of p-value, estimate and method used

Details

The statistical test used depends on the variable specified. if variable is factor with multiple levels, use Kruskal-Wallis test if variable is factor with 2 levels, use Wilcoxon test if variable is continuous, use Wilcoxon test

See also

sle.score delaneau.test

Examples

# load iris data
data(iris)

# variable is factor with multiple levels
# use kruskal.test
sle.test( iris[,1:4], iris[,5] )
#> $p.value
#> [1] 1.604944e-08
#> 
#> $estimate
#> Kruskal-Wallis chi-squared 
#>                   35.89518 
#> 
#> $method
#> [1] "kruskal.test"
#> 

# variable is factor with 2 levels
# use wilcox.test
sle.test( iris[1:100,1:4], iris[1:100,5] )
#> $p.value
#> [1] 0.7329198
#> 
#> $estimate
#> [1] -0.0005423853
#> 
#> $method
#> [1] "wilcox.test"
#> 

# variable is continuous
# use cor.test with spearman
sle.test( iris[,1:4], iris[,1] )
#> $p.value
#> [1] 0.000519093
#> 
#> $estimate
#>       rho 
#> -0.280033 
#> 
#> $method
#> [1] "cor.test"
#>