A Time Series Segment Finding Motifs Based On Sliding Window Algorithm

A sliding window based time series segmentation shape topic search algorithm segmentation is the foundation of time series data mining, and the quality of segmentation directly affects the results of data mining. The sliding window algorithm has attracted much attention due to its ability to achieve online segmentation. This article proposes a sliding window based time series segmentation algorithm. In response to the shortcomings of traditional sliding window algorithms that use a single threshold as the segmentation basis, this algorithm uses the number of data points within the window to adjust the fitting error and achieve dynamic adjustment of the segmentation threshold, thereby truly reflecting the changes in the time series. Through extensive experiments on actual datasets, it has been verified that this segmentation algorithm improves the accuracy of the algorithm while ensuring compression ratio.