Statistical Sequential Testing and Change-point Detection on Nonstationarity

2014 Fiscal Year Final Research Report

• Project/Area Number: 24530226
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Economic statistics
• Research Institution: Yokohama National University
• Principal Investigator: NAGAI KEIJI 横浜国立大学, 国際社会科学研究院, 教授 (50311866)
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 自己回帰過程 / 分枝過程 / 統計的逐次探索 / 変化点問題 / 停止時刻 / 統計的意思決定論 / 局所漸近正規 / 局所漸近最適性

Outline of Final Research Achievements

In this research, assuming online observations of stochastic sequence, I established a sequential testing procedure for the detection of the existence of non-stationary state such as bubble in financial time series and a sequential change-point detection procedure for the detection of the change in a parameter in financial time series. I also developed a sequential testing procedure and a sequential change-point detection for the criticality of branching process, which is important in epidemiology and population statistics. In each case, some mathematical optimality and a method for numerical computation are developed.

Free Research Field

経済統計