Adjacency-based regularization for partially ranked data with non-ignorable missing

作     者：Nakamura, Kento Yano, Keisuke Komaki, Fumiyasu 

作者机构：Univ Tokyo Grad Sch Informat Sci & Technol Dept Math Informat Bunkyo Ku 7-3-1 Hongo Tokyo 1138656 Japan RIKEN Ctr Brain Sci 2-1 Hirosawa Wako Saitama 3510198 Japan 

出 版 物：《COMPUTATIONAL STATISTICS & DATA ANALYSIS》 (计算统计学与数据分析)

年 卷 期：2020年第145卷第0期

页      面：106905-000页

核心收录：

学科分类：08[工学] 0714[理学-统计学（可授理学、经济学学位）] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：JST CREST [JPMJCR1763] MEXT KAKENHI [16H06533, 19K20222] JSPS Grants-in-Aid for Scientific Research [16H06533, 19K20222] Funding Source: KAKEN 

主　　题：Alternating Direction Method of Multipliers Expectation-Maximization algorithms Kendall distances Mallows models 

摘      要：In analyzing ranked data, we often encounter situations in which data are partially ranked. Regarding partially ranked data as missing data, this paper addresses parameter estimation for partially ranked data under a (possibly) non-ignorable missing mechanism. We propose estimators for both complete rankings and missing mechanisms together with a simple estimation procedure. The proposed procedure leverages the structured regularization based on an adjacency structure behind partially ranked data as well as the Expectation-Maximization algorithm. The experimental results demonstrate that the proposed estimator works well under non-ignorable missing mechanisms. Published by Elsevier B.V.