CAISEr (version 1.0.17)

summary.CAISEr: summary.CAISEr

Description

S3 method for summarizing CAISEr objects output by run_experiment()). Input parameters test, alternative and sig.level can be used to override the ones used in the call to run_experiment().

Usage

# S3 method for CAISEr
summary(object, test = NULL, alternative = NULL, sig.level = NULL, ...)

Value

A list object is returned invisibly, containing the details of all tests performed as well as information on the total number of runs dedicated to each algorithm.

Arguments

object

list object of class CAISEr (generated by run_experiment())

test

type of test to be used ("t.test", "wilcoxon" or "binomial")

alternative

type of alternative hypothesis ("two.sided" or "less" or "greater"). See calc_instances() for details.

sig.level

desired family-wise significance level (alpha) for the experiment

...

other parameters to be passed down to specific summary functions (currently unused)

Examples

# Example using four dummy algorithms and 100 dummy instances.
# See [dummyalgo()] and [dummyinstance()] for details.
# Generating 4 dummy algorithms here, with means 15, 10, 30, 15 and standard
# deviations 2, 4, 6, 8.
algorithms <- mapply(FUN = function(i, m, s){
  list(FUN   = "dummyalgo",
       alias = paste0("algo", i),
       distribution.fun  = "rnorm",
       distribution.pars = list(mean = m, sd = s))},
  i = c(alg1 = 1, alg2 = 2, alg3 = 3, alg4 = 4),
  m = c(15, 10, 30, 15),
  s = c(2, 4, 6, 8),
  SIMPLIFY = FALSE)

# Generate 100 dummy instances with centered exponential distributions
instances <- lapply(1:100,
                    function(i) {rate <- runif(1, 1, 10)
                                 list(FUN   = "dummyinstance",
                                      alias = paste0("Inst.", i),
                                      distr = "rexp", rate = rate,
                                      bias  = -1 / rate)})

my.results <- run_experiment(instances, algorithms,
                             d = 1, se.max = .1,
                             power = .9, sig.level = .05,
                             power.target = "mean",
                             dif = "perc", comparisons = "all.vs.all",
                             seed = 1234, ncpus = 1)
summary(my.results)

# You can override some defaults if you want:
summary(my.results, test = "wilcoxon")