Description

A shapelet is a time series subsequence that is identified as being representative of class membership. Classification with shapelets offers several benefits over competing approaches. Firstly, shapelets are directly interpretable and can offer explanatory insights into the problem domain. Secondly, the shapelet classifier is more compact than many of the alternatives, and hence classifying new instances is faster. Thirdly, shapelets allow for the detection of phase-independent shape-based similarity of subsequences. This type of similarity is often hard to detect with algorithms based on whole series. We introduce a new time series primitive, time series shapelets, which addresses these limitations.

While dozens of techniques have been introduced, recent empirical evidence has strongly suggested that the simple nearest neighbor algorithm is very difficult to beat for most time series problems, especially for large-scale datasets. While this may be considered good news, given the simplicity of implementing the nearest neighbor algorithm, there are some negative consequences of this. First, the nearest neighbor algorithm requires storing and searching the entire dataset, resulting in a high time and space complexity that limits its applicability, especially on resource-limited sensors. Second, beyond mere classification accuracy, we often wish to gain some insight into the data and to make the classification result more explainable, which global characteristics of the nearest neighbor cannot provide. In this work we introduce a new time series primitive, time series shapelets, which addresses these limitations. Informally, shapelets are time series subsequences which are in some sense maximally representative of a class. We can use the distance to the shapelet, rather than the distance to the nearest neighbor to classify objects.