Similarity Measure Based on Incremental Warping Window for Time Series Data Mining

Abstract

A similarity measure is one of the most important tasks in the fields of time series data mining. Its quality often affects the efficiency and effectiveness of the related algorithms that need to measure the similarity between two time series in advance. Dynamic time warping is one of the most robust methods to compare one time series with another based on warping alignments. In this paper, the design of an incremental warping window is used to improve the performance of dynamic time warping. The incremental warping window is changeable for various time series with different lengths. Moreover, the improved dynamic time warping based on the novel window considers the recent alignments as much as possible, which indicates that the proposed method concentrates on more information of the recent data points than that of the previous data points. In addition, it is suitable for online similarity measure between data stream. The experimental evaluation shows that the proposed method is effective and efficient for time series mining. INDEX TERMS Dynamic time warping, similarity measure, time series data mining, incremental warping window, classification.

Fast Constrained Dynamic Time Warping for Similarity Measure of Time Series Data

“…SC-DTW, I-DTW, and fast DTW, which all reduce the computational complexity of the standard DTW, were implemented for classification performance comparison, and the standard DTW was also implemented. The parameter r , indicating the window percentage value of SC-DTW and I-DTW, was set as 0.1, which is a commonly used value [46], and the parameter b, indicating the starting length of the I-DTW window, was set to 0 as in [40]. In addition, {0, 1, • • • , 9, 10} was used as the parameter r of the fast DTW, indicating the additional regions around the estimated optimal warping path.…”

“…To overcome the high complexity limitation of standard DTW, two major approaches have been reported. The indexing [37] technique reduces the calling times of standard DTW algorithm, while constrained DTW [38]- [40] and data abstraction [41]- [43] techniques reduce the calculations of standard DTW. By using a lower bounding function, the indexing technique reduces the calling times of standard DTW algorithm and performs DTW operation only for the remaining time series.…”

“…Two algorithms have been developed as constrained DTW techniques that consider only limited alignments. The Sakoe-Chiba DTW (SC-DTW) [38] performs equal limited alignments on all data points, while incremental DTW (I-DTW) [40] performs more alignments on recent data points than that on previous data points. These constrained DTW techniques can reduce the number of alignments to be considered and prevent the pathological alignment problem, which can reduce the overall classification accuracy due to excessive alignment between time series.…”

Arching Detection Method of Slab Track in High-Speed Railway Based on Track Geometry Data

“…By shifting initial aligned segments, the initial aligned sequence can be obtained (3) DTW-based secondary alignment DTW is a matching method for similar but warping sequences, which is essentially a dynamic programming problem [51]. It can search the optimal alignment relations between sequences efficiently and has been widely used to process series data [52,53]. Considering the spatial distortion of inspection data, the initial aligned segment Yi * is locally scaled based on DTW so that the mileage offset can be further corrected.…”

Fault diagnosis and identification of malfunctioning protection devices in a power system via time series similarity matching

“…Similarity matching is an important method for examining the analysis of time series data. In [17], the incremental warping window is used to improve the accuracy of matching among time series.…”