NEWS.md

In andjar/RMASCA: Assorted, Linear ASCA Functions

ALASCA 1.0.17

• Minor bug fix related to 1.0.16

ALASCA 1.0.16

• Bug fix: Plots show wrong number of variables when compared to n_limit for odd numbers
• Solves issue https://github.com/andjar/ALASCA/issues/12

ALASCA 1.0.14

• New function: predict_scores(). It accepts a data table with columns variable and value, and returns a new data table with a score column

ALASCA 1.0.11

• New features:
• optimize_PCs = TRUE (default: FALSE) will check if significant principal components have to be re-ordered during bootstrapping. This may happen if PCs are close in explanatory value so that they are shuffled during bootstrap. If this happens, it will trigger a warning
• waterfall = TRUE (default: FALSE) in plot() will replace points with bars for loadings. This can be very nice in combination with loading_group_column

ALASCA 1.0.10

• New feature: plot() now accepts the argument sort_loadings to control the order of the loading variables
• sort_loadings = "loading" sorts the variables by loading (default)
• sort_loadings = "alpha" sorts the variables alphabetically
• sort_loadings = c(...) sorts the variables in the same order as c(...), where ... is the variables of interest. Note that it may be required to increase n_limit (or use n_limit = 0) to ensure that all variables are shown. It is recommended to use n_limit = 0, i.e., plot(..., sort_loadings = c(...), n_limit = 0)

ALASCA 1.0.9

• Fix: Error when running LMMs without scaling

ALASCA 1.0.8

• Fix: Error when running linear models with Rfast (Error in crossprod(x, y) : requires numeric/complex matrix/vector arguments)

ALASCA 1.0.7

• Fix: Error for small datasets
• Fix: Error for custom stratification columns

ALASCA 1.0.5

• New feature: Black-and-white mode for more plot types
• Fix: Crash when combining use_Rfast = FALSE and another random intercept than ID
• Fix: Crash when trying to use only a three-way interaction as effect. Still a but unstable

ALASCA 1.0.4

• New feature: Plot effects in gray scale/black-and-white with symbols instead of colors. Can be tested with plot(..., bw = TRUE) (or grayscale = TRUE or greyscale = TRUE) or similarly with ALASCA(..., plot.bw = TRUE)
• Fix: Prediction plot with single-variable effect (e.g., time) did not color the groups correctly

ALASCA 1.0.3

• Fix: Crash when some participants are missing certain measurements

ALASCA 1.0.2

• New feature: Permutation testing (validation_method = "permutation")
• Simple permutation testing where data labels are shuffled at two levels: either the participant is re-assigned (e.g., a participant is randomly moved to a new group (or not)), or labels are shifted within participant (e.g. the time labels for a participant are shuffled)
• By default, the first effect is assumed to be shuffled within participant and the others shuffled across participant. The default can be overwritten by specifying permutation_within_participants and permutation_across_participants, e.g. permutation_within_participants = c("time")
• The participants should only belong to one group for each of the variables in permutation_across_participants and samples will be be reassigned as a block for permutation_within_participants (i.e., if a participant has two samples in group A and one sample in group B, then the two former samples will be reassigned together and not individually)
• Improved performance: df["value"][ rowNumers ] is somewhat faster than df[ rowNumers, value ]

ALASCA 1.0.1

• Fix: Error when using another column name than ID for ID

ALASCA 1.0.0

• First release with new framework

andjar/RMASCA documentation built on Sept. 15, 2024, 2:14 p.m.