Statistics for Big Data: Development of Theories and Tackling the 3Vs

2016 Fiscal Year Final Research Report

• Project/Area Number: 26540010
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Statistical science
• Research Institution: University of Tsukuba
• Principal Investigator: AOSHIMA Makoto 筑波大学, 数理物質系, 教授 (90246679)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: ビッグデータ / 潜在構造分析 / 異常値 / 欠損値 / 非正則推定論

Outline of Final Research Achievements

In this research project, we aim to pioneer new statistical theories for big data, ahead of the world. We have developed new theories and methodologies in latent structural analysis for big data: irregular and non-Gaussian data contaminated with outliers and missing values. New theories and methodologies guarantee stable and high accuracy at low computational cost. The findings of this research project are as follows: (1) Developments of the irregular inference theory for big data with diversity. (2) Developments of high-speed and highly accurate latent structural analysis for big data. (3) Pioneering latent structural analysis robust against outliers and missing values.

Free Research Field

統計科学