For smooth strictly increasing signals, the bias of isotonic regression is at most C(log n/n)^{β/3}, with matching lower bounds up to log factors; flat signals make bias as large as the error.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The bias of isotonic regression
For smooth strictly increasing signals, the bias of isotonic regression is at most C(log n/n)^{β/3}, with matching lower bounds up to log factors; flat signals make bias as large as the error.