• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2018 Fiscal Year Final Research Report

Variable selection methods in high-dimensional multivariate models and their applications

Research Project

  • PDF
Project/Area Number 16K00047
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Research Field Statistical science
Research InstitutionHiroshima University

Principal Investigator

Yasunori FUJIKOSHI  広島大学, 理学研究科, 名誉教授 (40033849)

Co-Investigator(Kenkyū-buntansha) 櫻井 哲朗  公立諏訪東京理科大学, 共通・マネジメント教育センター, 講師 (60609741)
Project Period (FY) 2016-04-01 – 2019-03-31
Keywords多変量回帰モデル / 変数選択法 / 主成分分析 / 判別分析 / 情報量規準 / 一つ取って置き法 / 高次元漸近枠組 / 多変量モデル
Outline of Final Research Achievements

In multivariate analysis, especially, discriminant analysis, multivariate regression models, principal component analysis, canonical correlation analysis, etc., we derived sufficient conditions for consistency of variable selection methods based on information criteria when the sample size and the dimension are large. The methods have a computational problem when the number of variables are large. In order to avoid its computational problem, we proposed a generalized kick-one-out (KOO) method which shared the high-consistency property in discriminant analysis and multivariate regression model. Further, in the problem of rank estimation in multivariate linear model, we proposed a regularized information criterion,
which can be used for the case that the sample size is smaller then the dimension of variables, and studied its high-dimensional properties.

Free Research Field

統計科学

Academic Significance and Societal Importance of the Research Achievements

多変量解析においては情報が入手しやすくなったこともあって, 変数の次元が大きい場合の統計的方法や, 膨大な変数の中から有用な変数を抽出する変数選択法に高い関心がよせられている. 本研究は, このような高次元データ分析における重要な課題における基礎的研究のみならず応用に焦点を当てており, 統計科学分野における理論や応用に関して高い貢献が期待される.

URL: 

Published: 2020-03-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi