A Time Series is a sequence T = (t1, t2,…,tn) which is an ordered set of n real-valued numbers. The ordering is typically temporal; however, other kinds of data such as color distributions (Hafner, Sawhney, Equitz, Flickner, & Niblack, 1995), shapes (Ueno, Xi, Keogh, & Lee, 2006), and spectrographs also have a well-defined ordering and can be fruitfully considered “time series” for the purposes of machine learning algorithms.
Motivation and Background
The special structure of time series produces unique challenges for machine learning researchers.
It is often the case that each individual time series object has a very high dimensionality. Whereas classic algorithms often assume a relatively low dimensionality (for example, a few dozen measurements such as “height, weight, blood sugar,” etc.), time series learning algorithms must be able to deal with dimensionalities in hundreds or thousands. The problems created by...
- Ding, H., Trajcevski, G., Scheuermann, P., Wang, X., & Keogh, E. A. (2008). Querying and mining of time series data: Experimental comparison of representations and distance measures. In Proceeding of the VLDB. VLDB Endowment.Google Scholar
- Ueno, K., Xi, X., Keogh, E., & Lee, D. (2006). Anytime classification using the nearest neighbor algorithm with applications to stream mining. In Proceedings of IEEE international conference on data mining (ICDM).Google Scholar