Full metadata record
DC FieldValueLanguage
dc.contributorDepartment of Computingen_US
dc.contributor.advisorYiu, Ken (COMP)en_US
dc.creatorNg, Yiu Wai-
dc.identifier.urihttps://theses.lib.polyu.edu.hk/handle/200/11361-
dc.languageEnglishen_US
dc.publisherHong Kong Polytechnic Universityen_US
dc.rightsAll rights reserveden_US
dc.titleSimilarity search for time series subsequence under dynamic time warpingen_US
dcterms.abstractTime series data can be pervasively found in many fields. We are always interested in retrieving knowledge from existing data. Finding similar subsequences could be the first step before preforming any analyses. Dynamic time warping is one of the best distance functions to measure similarity between two subsequences. However, the time complexity for computation of dynamic time warping is O(N2) for a query of length N, which is slow to calculate. To accelerate the computation, one method is to use lower bound functions. LB_Keogh and its variants are the most popular lower bound functions, which have time complexity of O(N). In this dissertation, we will focus on exact sequential search on normalized time series sequences under dynamic time warping for large dataset, with the assumption that dataset is non-segmented and lower bound function LB_Keogh is used. The state-of-art method is UCR Suite. The contribution of this dissertation is to improve the efficiency of UCR Suite under the scenario that (i) we are interested in finding similar subsequences for a few arbitrary-length queries, but not a single query, and (ii) the lengths of queries are long. A new lower bound function, namely LB_LowResED, is introduced. It is a lower bound function of LB_Keogh. The two assumptions are usually true for financial data analysis. Users would like to ask a few arbitrary-length long queries on the same set of financial data. This dissertation is composed of four parts: (i) introduction of time series sequences searching, (ii) related works, especially for the state-of-art method UCR Suite, (iii) introduction of the new lower bound function LB_LowResED, and (iv) experiment results.en_US
dcterms.extentvi, 40 pages : color illustrationsen_US
dcterms.isPartOfPolyU Electronic Thesesen_US
dcterms.issued2020en_US
dcterms.educationalLevelM.Sc.en_US
dcterms.educationalLevelAll Masteren_US
dcterms.LCSHTime-series analysis -- Data processingen_US
dcterms.LCSHHong Kong Polytechnic University -- Dissertationsen_US
dcterms.accessRightsrestricted accessen_US

Files in This Item:
File Description SizeFormat 
5809.pdfFor All Users (off-campus access for PolyU Staff & Students only)1.5 MBAdobe PDFView/Open


Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show simple item record

Please use this identifier to cite or link to this item: https://theses.lib.polyu.edu.hk/handle/200/11361