We use two different algorithms based on the way the sets A1, A2 are defined in (2). The first approach is called ‘‘left vs.
right’’ (LVR). If we divide the dataset into H slices, this approach uses the H − 1 cutoff points between the slices, denoted
as qr , r = 1, . . . , H − 1. Using the cutoff point qr we define ˜ Yr
LVR = I(Y > qr )