Title: Functional Data Analysis with the refund Package

Author: Philip Reiss

Affiliation: University of Haifa

 
Abstract:
The term functional data analysis (FDA) was introduced in the 1990s by Ramsay,
Silverman and colleagues to describe statistical methods for data sets in
which each observation is an entire curve or function, such as a time series
for each participant in an experiment. Many FDA methods are extensions of
classical statistical techniques, such as regression and principal component
analysis, from multivariate data to data that are curves or, more recently,
images. The FDA paradigm is well suited for intensive longitudinal data and
other complex data types encountered in psychology. 

This talk will introduce refund (REgression for FUNctional Data), a
collaboratively developed R package for FDA. Aside from methods for regression
with functional predictors and/or responses, the package also features several
approaches for principal component analysis with functional data. The talk
will include applications drawn from psychology and brain imaging.