A review on dynamic time warping in data visualization

Abstract

Dynamic Time Warping (DTW) is an algorithm for measuring the similarity between two time series that may vary in speed or length. Its ability to align sequences non-linearly makes it appropriate in a wide range of domains. However, its quadratic complexity and inability to handle data streams have been an obstacle to its adoption in modern data visualization. This paper explores the expansive scope of DTW within data visualization, focusing on its application in time series comparison. It introduces the fundamentals of DTW, explores the vast variety of modifications, and analyzes existing visualization techniques that uses DTW. Furthermore, since standard DTW is inadequate for modern visualization challenges (big data, streaming data, multidimensional data), it highlights the critical relevance of advanced DTW variants.