[R] Convex smoothing via 'Iterative Convex Minorant' ?

Martin Maechler maechler at stat.math.ethz.ch
Wed Jul 14 19:15:42 CEST 2004

I've been asked, and interested myself:

Has anybody implemented the above in R or another S language dialect?

We are talking about the algorithms / methodology
by Wellner, Groeneboom and Jongbloed, e.g., from the following article

  author =       "Geurt Jongbloed",
  title =        "The Iterative Convex Minorant Algorithm for
                 Nonparametric Estimation",
  journal =      j-J-COMPUT-GRAPH-STAT,
  volume =       "7",
  number =       "3",
  pages =        "310--321",
  month =        sep,
  year =         "1998",
  CODEN =        "????",
  ISSN =         "1061-8600",
  MRclass =      "62G05",
  MRnumber =     "1 646 718",
  bibdate =      "Sat Jan 2 17:33:21 MST 1999",
  URL =          "http://www.amstat.org/publications/jcgs/abstracts98/jongbloed.html",
  acknowledgement = ack-nhfb,


I know about Roger Koenker and Pin Ng's very interesting "nprq"
package (and its earlier version "cobs") where one can fit splines
with convexity constraints. 

At the moment however, we are specifically interested in the ICM
algorithm for convex nonparametric regression.

Thank you for pointers in advance.
Martin Maechler, ETH Zurich

More information about the R-help mailing list