fdr_envelope function

The FDR envelope

Calculate the FDR envelope based on the ATSE or IATSE algorithm of Mrkvička and Myllymäki (2023).

fdr_envelope(
  curve_sets,
  alpha = 0.05,
  alternative = c("two.sided", "less", "greater"),
  algorithm = c("IATSE", "ATSE"),
  lower = NULL,
  upper = NULL
)

Arguments

• curve_sets: A curve_set object or a list of curve_set

  objects containing a data function and simulated functions from which the envelope is to be constructed. Also envelope objects of spatstat are accepted instead of curve_set objects. If an envelope object is given, it must contain the summary functions from simulated patterns which can be achieved by setting savefuns = TRUE when calling the envelope function.

• alpha: The significance level. The 100(1-alpha)% global envelope will be calculated under the 'fwer' or 'fdr' control. If a vector of values is provided, the global envelopes are calculated for each value.

• alternative: A character string specifying the alternative hypothesis. Must be one of the following: "two.sided" (default), "less" or "greater". The last two options only available for types 'rank', 'erl', 'cont' and 'area'.

• algorithm: The algorithm for the computation of the FDR envelope. Either "IATSE" or "ATSE" standing for the iteratively adaptive two-stage envelope and the adaptive two-stage envelope, respectively, see Mrkvička and Myllymäki (2023).

• lower: A single number (or a vector of suitable length) giving a lower bound for the functions. Used only for the extension of the FDR envelope.

• upper: A single number (or a vector of suitable length) giving an upper bound for the functions. Used only for the extension of the FDR envelope.

Details

Typical use of this function is through other functions. fdr_envelope(cset) is the same as global_envelope_test(cset, typeone = "fdr"). Functions such as graph.fanova, graph.flm, frank.flm

allow to use the FDR control by specifying typeone = "fdr" appropriately (passing this to global_envelope_test).

Examples

# A GLM example
data(rimov)
nsim <- 1000 # Number of simulations

res <- graph.flm(nsim=nsim,
                 formula.full = Y~Year,
                 formula.reduced = Y~1,
                 curve_sets = list(Y=rimov),
                 factors = data.frame(Year = 1979:2014),
                 GET.args = list(typeone = "fdr"))
plot(res)

References

Mrkvička and Myllymäki (2023). False discovery rate envelopes. Statistics and Computing 33, 109. https://doi.org/10.1007/s11222-023-10275-7