Matrix Profile Evolution: An Initial Overview

Time series data have been investigated for decades in different domains. Recent fast development of wireless networks and cheaper price of small electronic monitoring devices, especially cheap IoT (internet of things) devices start to providing a lot time series data. However, those time series data are mixed with different patterns across lifetime. The patterns should be distinguished so the data can be separated and sent to corresponding process. There are different ways to tackle this challenge, for example by traditional pattern discovery or classification/clustering machine learning algorithms. The matrix profile (MP) method provides a way to handle this problem which can be used individually or together with other methods as an indicator variable or feature. This work aims to take an initial overview of MP method history and evolution from bibliometric aspect.