Big data analytics by multidimensional cluster scaling and its social applications

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330033
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Statistical science
• Research Institution: University of Tsukuba
• Principal Investigator: SATO-ILIC Mika (佐藤美佳) 筑波大学, システム情報系, 教授 (60269214)
• Research Collaborator: MARSALA Christophe University of Paris(UPMC), Department of Databases and Machine Learning, LIP6, 教授
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 分類 / ビックデータ / 尺度構成

Outline of Final Research Achievements

Multidimensional cluster scaling is developed as a novel method for big data analytics with an evaluation of its performance. The mainstream of the methodology for big data analytics includes the excluding data which has poor explainable power, reducing the data, and applying ordinary analytics. This methodology has a problem in relation to the validity of the result since the result depends on the criterion which determines the explainable power of the data. Therefore, in this study, the multidimensional cluster scaling was proposed in which all of the data information of the big data was used, and the data was analyzed in another space measured by another scale of the classification structure.

Free Research Field

統計科学