A Representation of Time Series Pattern Based on Skeleton Points
Download as PDF
Time series is the characteristic of a large amount of data, high dimension and fast data updating, so it is difficult to directly carry out data mining in the original time series. Drawing on the basic idea of linear segmentation of time series, a piecewise linear representation based on key points is proposed in this paper. Linear straight-line segments composed of skeleton points are used to approximate the time series. Skeleton points are regarded as the segmentation points to reflect the main features of time series. Experiment shows that this method can reduce the dimension of time series and minimize the overall error.
Time series, Algorithm, Piecewise, Skeleton points