STATISTIC AND APPLIED PROBABILITY SEMINAR
DATE and TIME: Nov 16-th, 2004 @1 pm in NS 234
SPEAKER: Ryan Gill (UofL Math Dept)
TITLE: Gentle Intro to Penalized Splines Theory
ABSTRACT: Penalized splines or P-splines are p-th degree spline
functions which
minimize the residual sum of squares with an additional penalty term
for jumps in
the p-th derivative at locations of the knots. These are similar
to
smoothing splines except the knots are placed at the sample
quantiles. This talk
will give a basic introduction to P splines and discuss selection of
the size of the penalty and the number of knots. A method based on
generalized cross validation (GCV) will be illustrated by an example
where we try to recover a sine curve based on simulated data with normal
random errors using P splines.