Statistical prediction, causation, incomplete data analysis and foundation of sciencee

点击翻译按钮获取中文摘要

Analysis of incomplete data has been troublesome both theoretically and practically. In particular nonignorable missingness has been a serious issue in statistics. An alternative perspective of the theory of missing data analysis is to provide an insightful view of statistical causal inference. Some notable research outcomes include development of the analysis of doubly censored data, a new method of exploring causal structure for data with latent confounders via the LiNGAM approach, incomplete data analysis with a shared-parameter model and development of the EM algorithm with constraints.