SemiPar is a package in the R language
for aiding semiparametric regression analyses,
and accompanies the book:
Ruppert, D., Wand, M. P. and Carroll, R.J. (2003).
Semiparametric Regression. Cambridge University Press.
Mid-2010 Stocktake of SemiPar (posted 4th May, 2010)
SemiPar 1.0-3 was posted on the
Comprehensive R Archive Network (CRAN)
in April 2010.
The SemiPar 1.0 Users' Manual:
explains what the package does.
Most of the code in SemiPar 1.0 runs as it should and works well.
However, since the initial CRAN posting in January 2005,
a few users have pointed out glitches with SemiPar 1.0.
It is difficult to find the large tracts of time
When fitting additive mixed models with a large number
of random effects
(usually corresponding to a large
number of subjects).
The cause seems to be that the
`Z matrix' becomes very large. The following script,
provides a remedy to this problem for the sitka spruce
data by avoiding creation of the contribution to the
Z matrix from the repeated measures.
Plotting can break down for higher-order
derivatives (e.g. plot(fit,drv=1)) in certain
The code for obtaining predictions also breaks down
in certain circumstances.
needed to make these fixes to SemiPar.
Meanwhile, the mgcv
package of Simon N. Wood (University of Bath, UK)
now does most of what we set out to achieve
when we started SemiPar in 1998 - and
is more rigorously maintained. So perhaps
it is better to leave SemiPar as a relic.