NEWS

datarium 0.2.0 (2026-07-23)

heartdisease: real, deliberately messy UCI four-site cardiology data (920 patients, 15 variables) with heavy, uneven missingness, for teaching data cleaning ahead of logistic regression. Redistributed under CC BY 4.0; see ?heartdisease.
infections: a simulated count outcome for teaching Poisson regression.
antismoking, taskachievment and renalstone: paired and repeated categorical data for McNemar's test, Cochran's Q test and the Cochran-Armitage trend test.
AirPassengersDf: the classic monthly international airline-passenger series (1949-1960) as a data frame, for time-series teaching.

Every data set now ships a full help page: a per-variable @format description (with units and factor levels), an honest @source (real vs simulated), @references, and a runnable base-R @examples that carries out the canonical analysis.
Every data set's help page now links to the Datanovia tutorial that teaches its analysis -- in the description and under "See also" -- and sibling data sets are cross-linked (mice/mice2, selfesteem/selfesteem2, headache/heartattack).
New "Meet the datarium datasets" article and a one-data-set-per-test map, available both in ?datarium and the README.
Package hex sticker (man/figures/logo.png).
Added a LICENSE.note documenting the data licensing: the author's data sets under CC0 1.0, heartdisease under CC BY 4.0.

The stress data set's id column is now stored as numeric (rather than integer) and some legacy column attributes were dropped; the values are unchanged.
Added data-raw/ provenance scripts documenting how each reproducible data set is generated.
Added a dataset-consistency check to R CMD check, backed by a byte-hash manifest that guards every shipped .rda.
Modernized the continuous-integration setup (R-CMD-check runs on the default branch).

Corrected factual errors in several data-set help files (dimensions and descriptive prose). Documentation only; no data values were changed.

First release