Recent Advances in Time-Series Data Mining: Similarity Measures & Representations

Abstract

In the last decade there has been an increasing interest in mining time series data since huge amounts are generated by several procedures in almost every domain such as in business, industry, medicine, science etc. Moreover, considering image or video data as time series data, the list of time series databases that need to be mined is expanded. During this period of time, hundreds of papers have been published covering all aspects of time series data mining, namely, dimensionality reduction or representation techniques, indexing, clustering, classification, novelty detection, motif discovery etc. Most of the contributions focus on proposing different dimensionality reduction approaches and providing novel similarity measures in order to deal with the unique characteristics of time series data, specifically, the high dimensionality, the high feature correlation and the large amounts of noise and to improve the performance of the existing data mining techniques. The objective of this paper is to serve as an overview of the most recent advances in the field of time series data mining. Although a general overview is included, the literature review is focused mainly on papers of the last three years.