fairy 0.6.5

results.html of the multinomineq analysis: the labels of the probabilities of unpacked models are no longer stretched; tables whose first column holds the numbers are centred in every column.

fairy 0.6.4

results.html of the multinomineq analysis: the tables use the full width and the trace and density plots are centred.

fairy 0.6.3

results.html of the multinomineq analysis: every second row of the tables is lightly shaded. The note for a model with an estimate of 0 now says that more samples may give a number.

fairy 0.6.2

results.html of the multinomineq analysis: the number columns of the tables (from the second column on) are centred; only the model name stays on the left.

fairy 0.6.1

R package fairy 0.6.1 (folder pkg).

Install:

remotes::install_github("marc-jekel/fairy", subdir = "pkg")

Changes since 0.6.0

• The analysis script of fairy_write_multinomineq() / fairy_multinomineq_script() and fairy_analyze_multinomineq() now also writes results.pdf: the settings, the Bayes factors of all models (coloured by the strength of evidence, after Jeffreys, 1961) and, for every model, the posterior (mean, SD, 95% interval, effective sample size, Geweke z) with trace and density plots.
• results.html: the legend of the evidence categories is one coloured scale (from "< 1/100" to "> 100"); the column "Representation" is left out when only the H-representation was analysed, which is what the app and the package do.
• The analysis from the V-representation (V_<model>.csv) is still prepared in the script, but switched off (commented out, with a note on how to switch it on); the files of the package contain the H-representation (A_, b_) only.
• Help of the prior argument shortened.

The full history is in pkg/NEWS.md.