Bag of Symbols for Time Series Distance Measurement and Applications
Pejman Khadivi · 2020
To have a precise understanding of the dynamics of a temporal dataset, it is of crucial importance to discover all the inter-relationships among the corresponding time series. In this paper, we propose a new distance metric to measure the mutual distance of time series in a temporal dataset. The proposed method, which works based on the concept of frequent bag of symbols and Jaccard index, is evaluated using a wide range of experiments with real and synthetic datasets. Experimental results show that the proposed method is superior over the baseline approaches from the performance and running time perspectives. Various real-world datasets, including COVID-19, climate change, and air pollution time series are used in this paper to demonstrate how the proposed approach performs in different big data and machine learning applications.