I4TSRS: A System to Assist a Data Engineer in Time-Series Dimensionality Reduction in Industry 4.0 Scenarios.

The massive captured data from industrial sensors (time-series data) that could serve as relevant indicators for predictive maintenance of equipment, fault diagnosis, etc. is generating a problem related to the considerable costs associated with their storage. In this paper we present a system called I4TSRS1, available as a Web Application, that efficiently guides a data engineer in the task of obtaining industrial time-series data reduced representations that preserve their main characteristics. Dealing with those reduced representations, data storage and transmission costs can be decreased, without limiting the future exploitation of the data in different processes. The novel contribution of the I4TSRS is that it is an intelligent system that recommends the best techniques to achieve a reduced representation of time-series captured in industrial settings. Its core element is a machine learning model that combines time-series reduction techniques with extracted features from industrial time-series. We have built the model using several heterogeneous real industrial time-series.