Segmentation of Time Series Data

View Sample PDF

Author(s): Parvathi Chundi (University of Nebraska at Omaha, USA)and Daniel J. Rosenkrantz (University of Albany, SUNY, USA)
Copyright: 2009
Pages: 6
Source title: Encyclopedia of Data Warehousing and Mining, Second Edition
Source Author(s)/Editor(s): John Wang (Montclair State University, USA)
DOI: 10.4018/978-1-60566-010-3.ch267

Purchase

View Segmentation of Time Series Data on the publisher's website for pricing and purchasing information.

Abstract

Time series data is usually generated by measuring and monitoring applications, and accounts for a large fraction of the data available for analysis purposes. A time series is typically a sequence of values that represent the state of a variable over time. Each value of the variable might be a simple value, or might have a composite structure, such as a vector of values. Time series data can be collected about natural phenomena, such as the amount of rainfall in a geographical region, or about a human activity, such as the number of shares of GoogleTM stock sold each day. Time series data is typically used for predicting future behavior from historical performance. However, a time series often needs further processing to discover the structure and properties of the recorded variable, thereby facilitating the understanding of past behavior and prediction of future behavior. Segmentation of a given time series is often used to compactly represent the time series (Gionis & Mannila, 2005), to reduce noise, and to serve as a high-level representation of the data (Das, Lin, Mannila, Renganathan & Smyth, 1998; Keogh & Kasetty, 2003). Data mining of a segmentation of a time series, rather than the original time series itself, has been used to facilitate discovering structure in the data, and finding various kinds of information, such as abrupt changes in the model underlying the time series (Duncan & Bryant, 1996; Keogh & Kasetty, 2003), event detection (Guralnik & Srivastava, 1999), etc. The rest of this chapter is organized as follows. The section on Background gives an overview of the time series segmentation problem and solutions. This section is followed by a Main Focus section where details of the tasks involved in segmenting a given time series and a few sample applications are discussed. Then, the Future Trends section presents some of the current research trends in time series segmentation and the Conclusion section concludes the chapter. Several important terms and their definitions are also included at the end of the chapter.

The IRMA Community

Research IRM

Segmentation of Time Series Data

Purchase

Abstract

Related Content

IRMA Sponsors