The returned ranking object of rankThenAggregate contains different variables holding the result of the aggregation depending on the data set.
For single-task challenges the name is rank_mean (if mean is chosen as aggregation function). For multi-task challenges the name is rank_FUN (independent from the chosen aggregation function). In aggregateThenRank the variable is named value_FUN for both single- and multi-task challenges.
The variables should be named consistently to avoid confusion.
Did you take care that FUN="mean" and FUN=mean is handled differently (in the former case it is a name, in the latter case it is a function)?
probably
now
e.g. if FUN="mean" (i.e. a character) or if FUN=mean (i.e. a function)
the column name will be "value_mean"
e.g. if FUN=function(x) quantile(x, prob = 0.5)
the column name will be "value_quantile(x, prob = 0.5)", i.e. carrying the function name.
@eisenman several tests unrelated to the fix fail (everything with "(reverse order)" is in the wrong order, which surprises me, should have failed already before, but I am unsure about this) could you please check?
The tests in test-rankThenAggregate.R are failing because the expectation is "rank_FUN" but should be "rank_mean" or "rank_median" respectively after the changes. This will be fixed in T27694.