Clustering Time Series Online In a Transformed Space
No Thumbnail Available
Authors
Hamid R Arabnia
Junfeng Qu
Yinglei Song
Khaled Rasheed
Byron A. Jeff
Issue Date
Type
Journal Article, Professional Journal
Language
Keywords
Alternative Title
Abstract
Similarity-based retrieval has attracted an increasing amount of attention in recent years. Although there are many different approaches, most are based on a common premise of dimensionality reduction and spatial access methods. Relative change of the time series data provides more meaning and insight view of problem domain. This paper presents our efforts to consider these relative changes of time series in similarity matching process. We propose a similarity distance measure that is based a series of critical points from the transformed difference space. Based on experiments with financial time series data, we conclude that our distance measure works as good as the Euclidean distance measure and PAA approach with normalized data. The proposed distance measure is a general distance metric therefore is suitable to deal with online similarity matching and it is not nessessary to maintain stream statistics over data streams during clustering of time series