A Preliminary Study on Multivariate Time Series Clustering

Abstract

© 2020, Springer Nature Switzerland AG. Time Series (TS) clustering is one of the most effervescent research fields due to the Big Data and the IoT explosion. The problem gets more challenging if we consider the multivariate TS. In the field of Business and Management, multivariate TS are becoming more and more interesting as they allow to match events the co-occur in time but that is hardly noticeable. In this study, Recurrent Neural Networks and transfer learning have been used to analyze each example, measuring similarities between variables. All the results are finally aggregated to create an adjacency matrix that allows extracting the groups. Proof-of-concept experimentation has been included, showing that the solution might be valid after several improvements.