optimal_group: Obtain optimal number of groups and degrees of freedom

In mshasan/empOPW: Empirical Method of Optimal Pvalue Weighting

Description

A function to obtain the optimal number of groups and the spline degrees of freedom from the p-values and the filters by applying the smooting spline regression

Usage

optimal_group(group = 5L, pvalue, filter, h_breaks = 71L, m, m1,
  alpha = 0.05, mean_testEffect, effectType = c("continuous", "binary"),
  method = c("BH", "BON"))

Arguments

group	number of groups to be used to split the p-values, default is five
pvalue	a vector of pvalues of the test statistics
filter	a vector of filter statistics
h_breaks	number of breaks to be used for the histogram, default is 71
m	total number of tests
m1	number of true alternatve tests
alpha	significance level of the hypothesis test
mean_testEffect	mean test effect of the true alterantives
effectType	type of effect sizes; c("continuous", "binary")
method	type of methods is used to obtain the results; c("BH", "BON"), Benjemini-Hochberg or Bonferroni

Details

Optimal group and degrees of freedom of the spline regression is a vital parameter to maximize the number of rejections. This function uses smooting spline regresion to obtain that.

Value

the number of rejected tests and the corresponding Optimal number of groups and the degrees of freedom of the spline smo0thing

Author(s)

Mohamad S. Hasan

Examples

# generate pvalues and filter statistics
m = 10000
set.seed(3)
filters = runif(m, min = 0, max = 2.5)          # filter statistics
H = rbinom(m, size = 1, prob = 0.1)             # hypothesis true or false
tests = rnorm(m, mean = H * filters)            # Z-score
pvals = 1 - pnorm(tests)                        # pvalue

results <- optimal_group(group = 10, pvalue = pvals, filter = filters,
             h_breaks = 71, m = m, m1 = 8000, alpha = .05,
             mean_testEffect = 2.5, effectType = "continuous", method = "BH")

mshasan/empOPW documentation built on March 1, 2021, 4:19 a.m.