Jukka Kohonen's SiZer implementation

This is my independent implementation of P. Chaudhuri and J. S. Marron's SiZer method. I wrote it as a part of my M.Sc. work titled Splinisilottimen käyttö SiZer-menetelmässä ("Using smoothing splines with the SiZer method").

Key features:

generates smoother family plots and SiZer maps;
two smoothing methods: smoothing splines and local linear regression;
three levels of inference: pointwise, row-wise simultaneous, and fully simultaneous inference;
two methods for doing simultaneous inference: multivariate Gaussian approximated with the "independent blocks" method, and bootstrap.

The library consists of a collection of MATLAB scripts. The most important ones are:

family, which plots a family of five smoothers at different smoothing levels;
sizer, which plots a SiZer map; and
combi, which does both.

Currently the only documentation, besides this file and my M.Sc. thesis (in Finnish), are the embedded comments in the scripts. Try

    help combi

in MATLAB.

Caveat: Please note that this software is provided "as is". There are known limitations, and there may be hidden bugs and inefficiencies. However, for experimental data analysis, it should be reasonably robust and easy to use.

Download the library (ZIP archive, 14 kB)