This R package is part of my Bachelor's Thesis which is supervised by the authors of the respective Papers (Prof. Dr. Markus Pauly & Dr. Paavo Sattler). My work has yet to be evaluated!
R package for performing tests on high dimensional repeated measure data for one group [1] or multiple groups [2].
The current version can be installed with:
## install devtools package
if (!requireNamespace("devtools", quietly = TRUE)) {
install.packages("devtools")
}
# install package
devtools::install_github("Schnieboli/hdrm", dependencies = TRUE)Be aware that the respective rtools-version for your version of R is required for installation.
A one group test can be performed by using the function
hdrm_single_longtabe or hdrm_single_widetable depending the format
of your data. Both take a data.frame as their first argument. The
package has two data sets included: birthrates [3] comes in a wide
table format and contains the birthrates of German states from 1990 to
2023. EEG [4] contains EEG data in 40 dimensions and comes in a long
table format.
library(hdrm)
### One sample test for data in wide table format with built in data set birthrates
hdrm_single_widetable(data = birthrates,
hypothesis = "flat", # test whether time profile is flat
)
### One sample test for data in long table format with built in data set EEG
# hypothesis can also be given as a matrix
hypothesis <- matrix(1/40, nrow = 40, ncol = 40) # matrix equivalent to 'flat'
hdrm_single_longtable(data = EEG,
hypothesis = hypothesis,
value = "value", # can all be given as a character...
subject = 4, # ...or a number
dimension = "dimension"
)A test for multiple groups can be performed by using the function
hdrm_grouped_longtabe or hdrm_grouped_widetable depending the format
of your data. As for the one group test, both take a data.frame as
their first argument. The test is performed using bootstraps to estimate
the computationally heaviest estimator. The number of bootstraps can be
controlled via the argument B. If your data is large, setting
bootstrap = TRUE will call the bootstrap versions for the other
estimators as well.
library(hdrm)
### Test for multiple groups for data in wide table format with built in data set birthrates
## a vector with the same length as ncol(data) is needed that divides subjects into groups
# divide German states in east and west:
group <- factor(c(1,1,2,2,1,1,1,2,1,1,1,1,2,2,1,2), labels = c("west","east"))
hdrm_grouped_widetable(data = birthrates,
hypothesis = "interaction", # test for interaction effect between group and dimension
group = group
)
### Test for multiple groups for data in long table format with built in data set EEG
# hypothesis can also be given as a matrix
hypothesis <- list(TW = diag(4) - matrix(1/4, 4, 4),
TS =matrix(1/40, 40, 40)
) # list entries equivalent to 'whole'
hdrm_grouped_longtable(data = EEG,
hypothesis = hypothesis, # test for time effect in groups
group = "group",
value = "value", # can all be given as a character...
subject = 4, # ...or a number
dimension = "dimension"
)[1] Pauly M, Ellenberger D and Brunner E 2015 Analysis of high-dimensional one group repeated measures designs Statistics 49 1243–61
[2] Sattler P and Pauly M 2018 Inference for high-dimensional split-plot-designs: A unified approach for small to large numbers of factor levels Electronic Journal of Statistics 12
[3] Statistisches Bundesamt (Destatis) 2024 Statistischer Bericht - Geburten 2023, Tabelle 12612-09 (Statisitsches Bundesamt)
[4] Höller Y, Bathke A C, Uhl A, Strobl N, Lang A, Bergmann J, Nardone R, Rossini F, Zauner H, Kirschner M, Jahanbekam A, Trinka E and Staffen W 2017 Combining SPECT and quantitative EEG analysis for the automated differential diagnosis of disorders with amnestic symptoms Front. Aging Neurosci. 9 290