Polynomial algorithms for isotonic regression
Victor D. Chepoi, Daniel Cogneau, Bernard Fichet · Lecture notes-monograph series · 1997
In this paper we consider the following problem.Let X = {#i,..., x n } be a set of observation points endowed with a partial order - R (i.e.Xi -< Xj implies that f( χ ί) < f( x j)) that minimizes the Z p -error We recall some general algorithms for solving this and related regression problems and we present new polynomial algorithms for some versions of the isotonic regression problem.