Graph-based prediction of missing KPIs through optimization and random forests for KPI systems

Marvin Carl May, Zeyu Fang, Michael B.M. Eitel, Nicole Stricker, Debarghya Ghoshdastidar, Gisela Lanza

Research output: Contribution to journalArticlepeer-review

4 Scopus citations

Abstract

Key performance indicators (KPIs) are widely used to monitor and control the production in industry. On an aggregated level, often represented as graphs or interrelated KPI systems, a comprehensive overview is given. However, missing or inaccurate sensor data and KPIs, as well inconsistencies in KPI based management are a major hurdle disturbing operations. To counter the impact of such missing KPIs, we propose a value optimization based approach to reconstruct the values of missing KPIs within a KPI system. While the approach shows successful reconstruction in the case study, the value optimization can be sped up through a random forest prediction of the initial optimization set. Thus, the inclusion of previous knowledge about the system behavior proves beneficial and superior to the pure optimization based approach, as validated by both randomized and simulation-based measurement data.

Original languageEnglish
Pages (from-to)211-222
Number of pages12
JournalProduction Engineering
Volume17
Issue number2
DOIs
StatePublished - Apr 2023

Keywords

  • Graph machine learning
  • KPI System
  • KPI prediction
  • Optimization

Fingerprint

Dive into the research topics of 'Graph-based prediction of missing KPIs through optimization and random forests for KPI systems'. Together they form a unique fingerprint.

Cite this