Title: Functional classification with direct and indirect effects for high dimensional data
Authors: Yuko Araki - Shizuoka University (Japan) [presenting]
Abstract: Recent years have seen that functional data analysis are capable of extracting intrinsic features from recently arising complicated and high dimensional data, such as three dimensional brain sMRI, hundreds of records of human gait, or traffic flow data for example. We introduce statistical methods for solving a classification problem which contains complex associations among several variables including high dimensional intermediate variables. The proposed method is based on composite basis function, which is an extended version of basis expansions with the help of sparse PCA. Further, $L_1$-type penalty constraints are imposed in estimation. This two-step regularization method accomplishes both covariates selection and estimation of unknown model parameters simultaneously. The crucial issue is how to select the regularization parameters used in model estimation. We propose a model selection method based information criterion. The proposed models are evaluated through Monte Carlo simulations to examine the efficiency of our modeling strategies.