3a ZDM Big Data Analytics: Environment

ZDM Data Analytics enables an efficient processing of industrial data in order to understand the behavior of the underlying system. It is a novel approach, specifically developed for the quality control in multi-stage production systems.

The entry barrier is low since there is no need for any a priori modelling from the data analytics point of view. The approach is based on unsupervised learning (of anomalies/outliers) and the only pre-requisite is the existence of data.

The main technological innovation is related to a novel method for data-driven detection of unusualities anomalies (variations) as outlier clusters in a multidimensional data space, whereas the metrics for defining clusters (i.e. calculating distance required for clustering) will be tailored to the nature of the analyzed data.

Information

• Paper: Nenad Stojanovic, Marko Dinic, Ljiljana Stojanovic: A data-driven approach for multivariate contextualized anomaly detection: Industry use case. IEEE BigData 2017: 1560-1569. (Access)