学术报告(Yihao Zhang)

题目：Element Oriented Analysis Aided Short-Term Prediction of Time Series
时间：12月23日， 10:30-11:30
地点：蒙民伟楼404会议室
 
报告人：
Mr. Yihao Zhang
PhD Candidate
Computing Department, ICS Division, Macquarie University

摘要：
Short term prediction of time series is one of the essential targets in tempor
al data mining. As a novel attempt, Element Oriented Analysis proposes to anal
yze time series from internal elements instead of data values themselves. In E
lement Oriented Analysis, we structure two consecutive levels to gradually obt
ain internal elements and then implement prediction. The task in the first lev
el is to find a precise change trend and detect a change function of time seri
es. The second level integrates the change function and selected lag variables
 into a predictive function and achieves time series prediction. The feasibili
ty of our approach is demonstrated in this paper by an empirical study on the
prediction of “Stock Index” from eight major stock markets.
 
简要
Yihao is currently a phd candidate at Macquarie University, Australia. He is n
ominated to implement the ARC Linkage Project (LP0561985). This project is pro
posed by The Australian Taxation Office (ATO) and Macquarie University togethe
r, which researches for the optimization of tax collections by improving the p
rediction and classification of taxpayer debt payment risk. Predictive models
for financial distress and insolvency risk will be developed in this project.
These models will be used to contribute to the understanding of taxpayers’ ci
rcumstances and the choice of appropriate collection actions.