Model-Driven Time-Series Analytics

Authors

  • Sabine Wolny CDL-MINT, TU Wien, Austria
  • Alexandra Mazak CDL-MINT, TU Wien, Austria
  • Manuel Wimmer CDL-MINT, TU Wien, Austria
  • Rafael Konlechner CDL-MINT, TU Wien, Austria
  • Gerti Kappel CDL-MINT, TU Wien, Austria

DOI:

https://doi.org/10.18417/emisa.si.hcm.19

Keywords:

Model-Driven Engineering, Time-Series, Data Analytics, Language Engineering

Abstract

Tackling the challenge of managing the full life-cycle of systems requires a well-defined mix of approaches. While in the early phases model-driven approaches are frequently used to design systems, in the later phases data-driven approaches are used to reason on different key performance indicators of systems under operation. This immediately poses the question how operational data can be mapped back to design models to evaluate existing designs and to reason about future re-designs. In this paper, we present a novel approach for harmonizing model-driven and data-driven approaches. In particular, we introduce an architecture for time-series data management to analyse runtime properties of systems which is derived from design models. Having this systematic generation of time-series data management opens the door to analyse data through design models. We show how such data analytics is specified for modelling languages using standard metamodelling techniques and technologies.

Downloads

Published

2018-02-27

Issue

Section

Invited Contribution