A0480
Title: Forecasting density-valued functional panel data
Authors: Cristian Felipe Jimenez Varon - King Abdullah University of Science and Technology (Saudi Arabia)
Ying Sun - KAUST (Saudi Arabia)
Han Lin Shang - Macquarie University (Australia) [presenting]
Abstract: A statistical method is introduced for modeling and forecasting functional panel data, where each element is a density. Density functions are nonnegative and have a constrained integral and thus do not constitute a linear vector space. A center log-ratio transformation is implemented to transform densities into unconstrained functions. These functions exhibit cross-sectionally correlation and temporal dependence. Via a functional analysis of variance decomposition, the unconstrained functional panel data is decomposed into a deterministic trend component and a time-varying residual component. A functional time series forecasting method based on the estimation of the long-range covariance is implemented to produce forecasts for the time-varying component. By combining the forecasts of the time-varying residual component with the deterministic trend component, h-step-ahead forecast curves are obtained for multiple populations. Illustrated by age- and sex-specific life-table death counts in the United States, the proposed method is applied to generate forecasts of the life-table death counts for 51 states.
