MultiTestH0 {DynClust}

Statistical test of zero mean for dynamics

Description

MultiTestH0 tests if each column vectors of a matrix seen as a noisy dynamic is of zero mean (H_0) or not. The multiple statistical test assumes known variance and is based on a multiple \chi^2 test.

Usage

MultiTestH0(proj.matrix, data.var, thrs = NULL)

Arguments

proj.matrix

a matrix whose colummns are tested to have zero mean or not.

data.var

a numeric providing the known variance

thrs

a numeric vector of thresholds specified as the 1-\alpha quantiles of the multiple test under the null on each partition. If thrs=NULL (default) return the global p-value which is the minimum of all p-values obtained on each partition.

Value

If thrs is provide returns a Boolean vector with length the number of columns of proj.matrix. Element j is TRUE if the null hypothesis (no difference with the null vector) is accepted for column j of proj.matrix. Otherwise, return one p-value per column.

Author(s)

Tiffany Lieury, Christophe Pouzat, Yves Rozenholc

References

Baraud Y., Huet S., Laurent B. Ann. Stat. (2003) Durot C., Rozenholc Y. Methods Math. Stat. (2006)