plot_raincloud_ly() gains scatter_box_gap (default box_width / 2 + 0.02) to set the distance from the group position to the scatter; it is now the same for every scatter_layout (#41).
plot_raincloud_ly() gains button_options to customise the position and style of the toggle buttons (#40).
plot_raincloud_ly()density_side and scatter_side deprecated; the density is always drawn right/top of the group position and the scatter left/bottom of it.
plot_raincloud_ly()scatter_layout option "dots" is renamed to "stacked", and "center" is removed; use scatter_width = 0 instead.
plot_raincloud_ly()jitter_width is renamed to scatter_width and now also sets the width of "stacked" scatter, so dot_spacing is removed.
Breaking: plot_corr_matrix() is renamed to plot_corrmat_ly(), so interactive plotly functions share the _ly suffix.
plot_corrmat_ly() hover text now shows the sample size n used for each pair, according to use (#33).
plot_corrmat_ly() with use = "na.or.complete" no longer errors when there are no complete rows; it returns NA correlations as stats::cor() does (use = "complete.obs" still errors).
Breaking: get_masked() no longer has a default threshold; it must now be specified (#37).
get_masked() now errors if vec is not numeric (including logical) or contains negative values and handles NAs; only values >= 0 and NA are accepted (#34).
get_masked() gains mask_na and mask_zero to allow user to choose whether whether to mask NAs or zeros (both default to FALSE, backward compatible). The function’s secondary masking now ensures at least two values are masked, at least one of them positive. Masked zeros and NAs count towards the two, but the smallest positive value(s) are always masked even if only zeros or NAs were masked (#34).
get_ultimate() gains na.rm (default TRUE, backward compatible). Set na.rm = FALSE to treat NA as a meaningful value considered as a valid candidate. exclude must not contain NA (it now errors); use na.rm instead (#35).
Fix typo in npdar-package.R.
npdar 0.6.0
Rewrite README.md with README.qmd and update .Rbuildignore.
Add light switch and MathJax for math notations in pkgdown site.
Breaking: renamed functions to snake_case; the domain families drop get_, the utilities keep it:
Edit help aliases bp (for bp_family) and fiscal_year (for audit_year_family).
Breaking: In bp_family, bp_assess() no longer takes option; it validates inputs once and returns a tibble with expected, zscore, percentile and category.
In bp_family, functions now error if sex, age_years, height_z and bp_value differ in length (no recycling), and bp_relative() returns one row per input for NICE/BHF.
In bp_family, fixed bp_category() and bp_assess() returning “Prehypertension” for BP below the lower bp_limit; these now return NA.
npdar 0.5.0
Added count_na argument to get_frequency() to allow user to count missing values in measures as their own category (category = NA) and include them in the denominator. It is backward-compatible as default is FALSE (exclude NAs) (#28).
get_frequency() now returns measures in the order specified in measures instead of alphabetical/numerical order. (#29).
Simplified get_frequency() and get_corrMat() internals and tidied the unit tests.