calculate.power: Calculate power at a given significance level

In knnIndep: Independence tests and benchmarks

Description

Function to calculate power at a given significance level. Uses the data structure returned by run.tests

Usage

calculate.power(vals, alpha = 0.95, comp = `>`)

Arguments

vals	list, values as returned by run.tests
alpha	significance level at which to return power
comp	comparison function, for alpha < .5, it should probably be set to '<'

Details

power is calculated as the fraction of tests that are higher or lower than (according to comp) than the significance level. The significance level is fixed on data generated under the null hypothesis.

Value

returns the power for applicable data from the structure vals, usually for each test it returns the power for all types of dependence and all noise levels.

Author(s)

Sebastian Dümcke [e-mail]

See Also

run.tests

Examples

mycor = function(...) cor(...)^2
vals = run.tests(mycor,list(),1:2,cbind(c(.3,.4,6),c(.3,.5,4)),100)
drop(calculate.power(vals))

knnIndep documentation built on May 2, 2019, 3:23 a.m.