A1066
Title: Functional data analysis
Authors: Dirong Chen - Beihang University (China) [presenting]
Abstract: In a random sample of functional data, each subject is recorded as one or several functions. The high or infinite dimensional structure of these data is a rich source of information. On the other hand, the high intrinsic dimensionality of the data poses challenges both for theory and computation. The purpose is to introduce some methods in functional data analysis, with emphasis on the estimation of mean functions and covariance functions based on discretely observed data. A data-driven approach is proposed based on the framelet block thresholding. It has the advantages of adaptivity to local spatial, global smoothness and sampling frequency.
