Variable selection problem for mixed effects model and its application to small area estimation

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K17101
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Economic statistics
• Research Institution: Chiba University
• Principal Investigator: Kawakubo Yuki 千葉大学, 大学院社会科学研究院, 准教授 (80771881)
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 小地域推定 / 情報量規準

Outline of Final Research Achievements

In this program, I mainly conducted the following two pieces of research in the variable selection problem for linear mixed model, which is used in small area estimation.
First, I considered the covariate shift situation, which is when the values of covariates in the model for prediction differ from those in the model for observed data. Under the covariate shift situation, I constructed a variable selection criterion for linear mixed model based on the conditional AIC. Second, I considered variable selection and estimation simultaneously to minimize the total mean squared prediction errors of the predictor. The derived method can be regarded as an extension of the observed best prediction (OBP) method.
I also conducted research related to small area estimation or mixed effects model.

Free Research Field

経済統計

Academic Significance and Societal Importance of the Research Achievements

本研究課題であげた研究成果は，線形混合モデルの変数選択における新しい手法を提案したものであり，また提案手法の理論的な性質についても議論している。ともに統計学分野の国際査読誌に掲載され，一定の学術的意義を認められたと言える。
またここであげた研究成果は，線形混合モデルを用いた統計手法に広く応用可能であるが，その中でも特に小地域推定と呼ばれる官庁統計における重要な課題への応用を強く意識して研究を行ってきた。小地域推定における実務での有用性という観点からも，社会的意義のある研究成果だと言える。