Testing the equality of two sample high dimensional mean vectors using the method developed in arXiv:1406.1939 [math.ST]
twoMeans(X, Y, m = 2500, filter = TRUE, SX = NULL, SY = NULL,
alpha = 0.05, DNAME)The n x p training data matrix.
The n x p training data matrix.
The number of repetition in the test, default to be 2500
A logical indicator of the filtering process, default to be TRUE
The covariance matrix of X, if not presented it will be estimated from the input sample.
The covariance matrix of T, if not presented it will be estimated from the input sample.
The significant level of the test.
Defaulf input.
Value of testing statistics, p-values (the non-studentized statistic and the studentized statistic respectively), alternative hypothesis, and the name of testing procedure.
Implement the method developed in arXiv:1406.1939 [math.ST] to test whether a high dimensional mean vector is zero or not, which is equivalent to test \(H_0: \mu_1=\mu_2\). The procedure utilizes bootstrap concept and derive the critical values using independent Gaussian vectors whose covariance is estimated using sample covariance matrix.
J. Chang, W. Zhou and W.-X. Zhou, Simulation-Based Hypothesis Testing of High Dimensional Means Under Covariance Heterogeneity (2014), arXiv:1406.1939.