A study of joint dimension reduction and clustering for heuristic considerations of large-scaled data

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330052
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: Okayama University of Science
• Principal Investigator: Mori Yuichi 岡山理科大学, 総合情報学部, 教授 (80230085)
• Co-Investigator(Renkei-kenkyūsha): Adachi Kohei 大阪大学, 人間科学研究科, 教授 (60299055) ; Nakano Junji 統計数理研究所, 統計計算開発センター, 教授 (60136281)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 多変量解析 / 次元縮約 / 非計量主成分分析 / 非計量因子分析 / クラスター分析 / 加速化アルゴリズム / 対話的可視化ツール

Outline of Final Research Achievements

The purpose of this study is to propose joint dimension reduction and clustering to deeply consider the features of data, particularly large-scales data. To do this, the followings are performed: gathering information in previous studies, reviewing dimension reduction (including variable selection), proposing dimension reduction and clustering dealing with mixed measurement level data, developing interactive interface, and improving computational efficiency. The study is successful to provide some methods that handle the complexity including mixture of measurement level in the context of principal components analysis, and that consider the features of data by heuristic thinking and trial and error using developed interactive interfaces and proposed acceleration algorithms in case where a huge number of computation is necessary such as large-scaled data processing and variable selection.

Free Research Field

計算機統計学