Marginal distribution clustering of multi-variate streaming IoT data

摘要

Analysis and processing of time-series data has been studied over the past decades and has resulted in several promising algorithms. Clustering continuous data is an interesting subset of unsupervised learning models to process and categorise the data. Over a long period and with a large number of samples most conventional data streams will converge to Gaussian distributions. The existing clustering methods for continuous data are usually suitable for this type of data. However, with growing volumes of real world observation and measurements, (i.e. Internet of Things data), the data sets become more volatile and have multi-variate distributions. In this paper we propose a dynamic and adaptable clustering algorithm for multi-variate time-series data clustering. We have evaluated our work against some well-known time-series clustering methods and have shown how the proposed method can reduce the complexity and perform efficient in multi-variate data streams.