Causality Mining from Event Sequence Data and Its Applications to Causality Discovery in Earthquakes and Damages

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K16052
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: Osaka University
• Principal Investigator: Fukui Ken-ichi 大阪大学, 産業科学研究所, 准教授 (80418772)
• Research Collaborator: Okada Yoshiyuki 大阪大学, 大学院情報科学研究科, 大学院生 ; Sato Kazuki 大阪大学, 大学院情報科学研究科, 大学院生
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: データマイニング / クラスタリング / 頻出パターン / ベイズ推定 / 弾性マッチング / 燃料電池 / 地震 / 発生相関

Outline of Final Research Achievements

In this research, we proposed a new data mining algorithm, called Cluster Sequence Mining (CSM), which extracts occurrence correlation between events from multidimensional event series data. Furthermore, we extended the correspondence relation when calculating the time interval between events to one-to-many or many-to-one, by formulating as a minimum cost elastic matching problem, and devised a method to uniquely obtain corresponding event pairs. This aimed at improving the accuracy of Bayesian estimation. As a result of the evaluation experiment using artificial data, accuracy improvement was confirmed in the proposed method, especially when the event densely exists on the time axis as compared with the conventional method. Furthermore, we showed examples of applying this method to damage correlation analysis of fuel cells and occurring correlation analysis between earthquakes.

Free Research Field

データマイニング