|
Abstract:
|
This talk introduces an imputation method for generating missing values of sparse functional data. The data consist of observations of smooth curves (e.g., growth curves) at sparse time points, and the problem is to find a way to carry out functional principal components analysis and functional regression modeling based on unobserved gradients (e.g., growth rates). Existing approaches are only appropriate when the aggregated time points become dense, and are not specifically tailored to gradients. We introduce a way of imputing the gradients that is consistent with the data, and that provides a natural and direct method of estimating their principal components and fitting functional linear regression models.
|
