to_result_table: Generate data frame with aggregated data and statistical test...
In jakobbossek/tblutils: Helpers to pimp scientific tables

Description Usage Arguments Value

View source: R/to_result_table.R

Often we are given a large data.frame with results of multiple (stochastic/randomized) algorithms on multiple instances with varying parameters. One common way to present such results is to aggregate over runs, e.g. on instance level, apply statistical singnificance tests to underpin observations and eventually to present these ggregated results by means of result tables. These result tables contain meta columns (e.g. problem, parameters) and each at least one column for each algorithm.

This function serves for aggregation and calculation of statistical tests. The result is a list of aggregated result tables which may be passed on to to_latex.

to_result_table(x, split.cols, widen.col, measure.cols,
  do.pairwise.test = TRUE, test.alternative,
  testresult.col = "testresult", alpha = 0.05,
  stats.formatter.args = list(sd = list(digits = 2L)),
  testresult.formatter.args = list(positive.only = TRUE),
  highlighter.args = NULL)

`x`	[`data.frame`] Input data frame.
`split.cols`	[`character`] Column name of columns used to split the data. Usually this will be something like “problem/instance”, “parameter”.
`widen.col`	[`character(1)`] Column used to widen the table, convert long to wide format. Usually, this is the “algorithm”.
`measure.cols`	[`character`] Which columns contain the actual measurements (e.g. performance values, number of iterations etc.). If multiple are provided and `do.pairwise.test` is `TRUE`, the data is additionally split by `measure.cols`.
`do.pairwise.test`	[`logical(1)`] Should pairwise test of measurements be performed? Default is `TRUE`.
`test.alternative`	[`character`] Alternatives, i.e. H_1-hypothesis, for statistical tests.
`testresult.col`	[`character`] Column name where test result shall be stored. Default is “testresult”.
`alpha`	[`numeric(1)`] Significance level for statistical tests. Default to `0.05`.
`stats.formatter.args`	[`list of lists`] Named list (names are aggregated measures) of arguments used to format the aggregated measurement.
`testresult.formatter.args`	[`list`] Named list of parameters passed down to LaTeX-formatter for test results.
`highlighter.pars`	[`list`] Named list of arguments for highlighter, i.e. function that is used to emphasize certain results in tables (e.g. the minimal mean value on each instance).