B1632
Title: Generalized functional linear mixed model
Authors: Ruvini Jayamaha - Western Michigan University (United States) [presenting]
Hyun Bin Kang - Western Michigan University (United States)
Abstract: Functional data analysis (FDA) has been an area of concern in scientific communities with the advancement in data collection technologies. Specifically, researchers in various fields such as anthropology, epidemiology, neurology and chemometrics face the challenges of obtaining useful information from more detailed, complex and structured data. Since the existing methods often are not suitable for such data, new statistical approaches are developed to accommodate these complicated data structures. In FDA, the fundamental statistical unit is a function or curve instead of a vector of measurements. The natural smoothness in the data can be exploited to achieve greater statistical efficiency compared to the multivariate statistical methods. A generalized functional linear mixed model (GFLMM) is presented, an extension of classical generalized linear mixed models to include the functional covariates. This framework considers the regression situation where the response variable is scalar, and the predictor is a random function. The situation where the link function and variance functions are unknown and are estimated nonparametrically from the data are also considered. A semiparametric quasi-likelihood procedure and the Monte Carlo method are used for the estimation and inference in GFLMM.
