There are three datasets loaded by data("skating")
:
skating
, a log-likelihood function for the competitors'
strengths, skating_table
, an order table for each of the 9
judges, and skating_maxp
, the result of maxp(skating)
,
which is included to save time in the examples.
These objects can be generated by running script
inst/skating_analysis.Rmd
, which includes some further
discussion and technical documentation. The dataset is interesting
because it has been analysed by many workers, including Lock and Lock,
for consistency between the judges.
Note that file is structured so that each competitor is a row, and
each judge is a column. Function rank_likelihood()
requires a
transpose of this to operate.
Object skating_table
is an order table, taken from Lock and
Lock. It corrects what appears to be an error in which judge 5 ranked
both Butyrskaya and Kettunen 12; there is no 13. Using EM, I reckon
that Butyrskaya should be ranked twelfth and Kettunen thirteenth.