www-ai.cs.tu-dortmund.de/LEHRE/SEMINARE/SS09/AKTARBEITENDESDM/LITERATUR/WrobelRuepingSubgroup.pdf
LNAI 5211 - Tight Optimistic Estimates for Fast Subgroup Discovery
one. Here is another one: consider p0 = (0.1, 0.45, 0.45) (for c = 3) respectively p0 = (0.1, 0.3, 0.3
c(3 , ..., 0.3 c(3 , 0.3) (for c > 3). Furthermore,
consider a subgroup s with m = (10, 10, 0, 0, .. [...] is not tight it is su"cient to note that np(1-p0) < n(1 - p0) for some n > 0, p0 < 1 and p < 1. !
3.3 Tight Estimates for the Multi-class Quality Functions Split, Gini and Pearson’s !2
Next, we turn to [...] calculated and used as an estimate.3 Here is the definition of the scheme oed
q of optimistic estimates, where d de- termines the number of classes considered at a time:
3 This scheme is a generalization …