Applying Trie-Structure to Improve Dynamic Time Warping on Time-Series Stock Data Analysis

摘要

Dynamic time warping (DTW) is a robust but time-consuming on distance measure for time-series data similarity search. To speed up DTW for time-series data analysis, a new approach with trie-structure is introduced for the process of the DTW recognition steps, such as in financial stock pattern analysis. The final result shows that searching time has been reduced by this approach while applying trie-structure to DTW on tune-series data analysis, especially on stock data.