computeMemory() when labeling subset rows for trend direction ("up"/"down"/"stable").computeMemory() to apply addRandomColumn() to the subsetted data (lagged.data.model) rather than the full dataset.plotExperiment() referencing wrong variable name (parameters instead of parameters.file).prepareLaggedData() using hardcoded "age" instead of the user-supplied time column name in time.zoom validation.runExperiment() operator precedence bug: length(parameters.names == 1) changed to length(parameters.names) == 1.importFrom directives into individual entries in NAMESPACE.num.threads to computeMemory() to allow multithreading in ranger::ranger().add.random from computeMemory() and replaced it with the option "none" in argument random.mode.response and drivers are no longer required in computeMemory() if argument lagged.data was generated with prepareLaggedData().ribbon to plotMemory() and removed other useless arguments to simplify the usage.sampling.names from runExperiment(), plotExperiment(), and experimentToTable() as it provided minimal value while adding complexity.sampling.subset from extractMemoryFeatures() as it provided minimal value while adding complexity. Users can filter input data directly (e.g., data[data$sampling == 25, ]).mergePalaeoData() to alignTimeSeries() for domain-agnostic naming. The old name remains as a deprecated alias.prepareLaggedData() to lagTimeSeries() for consistency with alignTimeSeries(). The old name remains as a deprecated alias.computeMemory(), extractMemoryFeatures(), mergePalaeoData(), plotMemory(), prepareLaggedData(), runExperiment(), and dataset help pages.plotInteraction(), as it is available in spatialRF::plot_response_surface() and the package pdp.Initial CRAN release.