Prof. Ramsay's Ftp Site

This site contains code for standard FDA functions, such as those used in analyses presented in workshops, with versions for use in Matlab, R, or SPLUS. These functions were developed in a Windows environment, and some changes may be required for Unix installations.

The Matlab code was last updated on March 28, 2007 to reflect changes in Version 7 of Matlab. Data and sample code of some of the workshop examples are also here, along with data and code for other examples found in Functional Data Analysis and Applied Functional Data Analysis texts.

The R and S-PLUS code was updated on M, to bring the code up to date with the Matlab code. Before this date, the functions and examples were about two years behind the Matlab code, so that it is certainly recommended that you update your version if you have not done so very recently.

An R package has been submitted to the R-Project for testing, and should replace the old package soon. We are especially grateful to Mr. Hadley Wickham, a grad student at Iowa State University, for his help in setting the package up.

The new R/S-PLUS functions use two new classes in addition to the three already introduced. The fdPar class defines functional parameter objects, which are functional data objects with additional slots defining the roughness penalty used to create them. The Lfd class defines linear differential operators in a more flexible way than previous versions. Along with changes in the functions, the examples have been extended to reflect the new approaches discussed in the second edition of Functional Data Analysis. A revised manual is also available along with the new functions.

The Matlab versions of these functions are often much faster than their R and S-PLUS counterparts, and especially for optimizations requiring iterative improvement of an objective function, such as for monotone and positive smoothing, and for continuous registration. The S-PLUS versions do use some C++ code for speed up computation, but the R versions do not at this date (16 April 2007). We are working on extending the FDA C++ code and incorporating it into the R and S-PLUS versions.