A0848
Title: Characterizing heterogeneous dynamics in multiple-subject multivariate time series
Authors: Zachary Fisher - Penn State University (United States) [presenting]
Abstract: Heterogeneity is a ubiquitous and defining feature of human behavior. At the same time, how best to characterize and intervene on heterogeneous processes remains a critical open question. Part of the difficulty in addressing heterogeneity lies in the fact that individuals differ from one another in complex and meaningful ways, and at the same time, dynamics within individuals themselves often evolve and adapt across time and context. Recently, a study introduced the multi-VAR framework for simultaneously modeling multiple-subject multivariate time series characterized by common and individualizing features using penalized estimation. This approach differs from many popular modeling approaches for multiple-subject time series in that both qualitative and quantitative differences in a large number of individual-level dynamics are well-accommodated. Extensions of the original multi-VAR approach are presented to accommodate nonstationary series by allowing individual-level dynamics to vary over time.
