Adaptive window based piecewise linear representation algorithm
Huiling Zhong · Shenyang Gongye Daxue xuebao · 2014
In order to reduce the fitting error of fixed window based piecewise linear representation algorithm, six kinds of data change patterns in the time series data were abstracted through the analysis. Therefore,the adaptive window was designed to load the different patterns of data and solve the segmentation point,and an adaptive window based piecewise linear representation( AW-PLR) algorithm was established. The results based on the real GPS floating car data and general test data show that the compression ratio and fitting error can be controlled through adjusting the threshold r with the AW-PLR algorithm. Under the same compression ratio, the AW-PLR algorithm can averagely reduce the fitting error by 24% ~ 27%,compared with the SEEP algorithm.