Cluster Analysis of Time Series Data Using the K-Means Method with Dynamic Time Warping.

Saved in:

Bibliographic Details
Title:	Cluster Analysis of Time Series Data Using the K-Means Method with Dynamic Time Warping.
Authors:	Saputro, Vito Christian Setiawan¹ vitochrs8@gmail.com, Robby² robby@unpar.ac.id, Sukmana, Agus² asukmana@unpar.ac.id
Source:	Engineering Letters. Jun2026, Vol. 34 Issue 6, p2274-2284. 11p.
Subjects:	K-means clustering, Cluster analysis (Statistics), Data transformations (Statistics), Stock price indexes
Abstract:	Stock clustering serves as a quantitative tool to support portfolio construction by enabling investors to identify groups of stocks that exhibit similar behavior. In this study, clustering is applied to stocks included in the LQ45 index based on similarities in their price movement patterns. The clustering procedure employs the K-Means algorithm combined with Dynamic Time Warping (DTW) as the distance measure. DTW is selected because of its ability to more effectively capture similarities in time series movement patterns compared to conventional distance metrics. Time series clustering using the K-Means framework requires the specification of representative cluster centroids. To this end, the DTW Barycenter Averaging (DBA) method is utilized to compute the centroid of each cluster. Empirical results indicate that data normalization leads to clusters that more accurately reflect underlying stock movement patterns than clustering performed on non-normalized data. The centroids obtained through the DBA method effectively characterize the common dynamics of stocks within each cluster. Furthermore, the findings suggest a relationship between correlation structure and movement pattern similarity, where higher intra-cluster correlations are associated with more homogeneous stock movement behaviors. [ABSTRACT FROM AUTHOR]
	Copyright of Engineering Letters is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Stock clustering serves as a quantitative tool to support portfolio construction by enabling investors to identify groups of stocks that exhibit similar behavior. In this study, clustering is applied to stocks included in the LQ45 index based on similarities in their price movement patterns. The clustering procedure employs the K-Means algorithm combined with Dynamic Time Warping (DTW) as the distance measure. DTW is selected because of its ability to more effectively capture similarities in time series movement patterns compared to conventional distance metrics. Time series clustering using the K-Means framework requires the specification of representative cluster centroids. To this end, the DTW Barycenter Averaging (DBA) method is utilized to compute the centroid of each cluster. Empirical results indicate that data normalization leads to clusters that more accurately reflect underlying stock movement patterns than clustering performed on non-normalized data. The centroids obtained through the DBA method effectively characterize the common dynamics of stocks within each cluster. Furthermore, the findings suggest a relationship between correlation structure and movement pattern similarity, where higher intra-cluster correlations are associated with more homogeneous stock movement behaviors. [ABSTRACT FROM AUTHOR]
ISSN:	1816093X