Accurate Time Series Classification Using Partial Dynamic Time Warping

CHOTIRAT ANN, RATANAMAHATANA and HAEMWAAN, SIVARAKS and PHONGSAKORN, SATHIANWIRIYAKHUN and THAPANAN, JANYALIKIT (2015) Accurate Time Series Classification Using Partial Dynamic Time Warping. In: Second International Conference on Advances in Applied Science and Environmental Technology - ASET 2015, 28-29 August, 2015, Bangkok, Thailand.

20150901_103238.pdf - Published Version

Download (672kB) | Preview
Official URL:


Dynamic Time Warping (DTW) has been widely used in time series domain as a distance function for similarity search. Several works have utilized DTW to improve the classification accuracy as it can deal with local time shiftings in time series data by non-linear warping. However, some types of time series data do have several segments that one segment should not be compared to others even though DTW can naturally warp across those segments. In this paper, we propose PartialDTW distance measure that utilizes domain knowledge about special characteristics of different sections of the data to limit the warping path. The experiment shows that our PartialDTW has much better performance when compare with other well known algorithms.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Dynamic time warping, DTW, PartialDTW, time series classification
Depositing User: Mr. John Steve
Date Deposited: 27 Apr 2019 06:34
Last Modified: 27 Apr 2019 06:34

Actions (login required)

View Item View Item