• Hero 1
  • Hero 3
  • Hero 2

Predicting the Future is a Matter of Trust

Thursday 2nd December 12:15 - 12:45

Prediction is very difficult, especially if it's about the future. Niels Bohr, the Nobel laureate in Physics and father of the atomic model, was right with his quotation, although he probably did not know the capabilities of Machine Learning in 2021.

Prediction is difficult because of its complexity. Looking at metals and aluminium downstream production, it is common knowledge that there are reams of parameters that influence the quality. It is not only each single value that has impact on the result, but also the interaction and the constellation of the parameters has a major effect on the outcome.

Machine Learning is an essential tool to resolve this complexity. The adoption of these techniques in the reality of the metals’ industry requires an insight into the results produced by Predictive Models. The maturity of the model should be confirmed thorough inspection, employing Machine Learning Interpretability techniques. On one hand, it provides a deep understanding of the predictive model behaviour under a variety of circumstances. On the other hand - in connection with the domain knowledge - it brings an efficient tool for Root Cause Analysis.

In our real use case applied in an aluminium rolling mill, the process data gathered during the aluminium coil production was used for the creation of Machine Learning models, where the predictive target was the occurrence of a defect. These models, subsequently, were exposed through the Machine Learning Interpretability techniques. We will show the application of these techniques and how to extract business value from certain aspects of Machine Learning interpretations.

We do not trust data we do not understand. Hence, our focus is on showing the application of these techniques and extracting business value from certain aspects of Machine Learning interpretations. A special emphasis will be placed on the prediction breakdown, which is a decomposition of a single prediction into the contributions from all involved predictors. This is a measure of their importance and provides precise information about the impact of a given data and process feature in the context of a particular prediction. All mentioned techniques are applied to understand the decisions proposed by the models. This provides confidence and trust that the predictions are fair and based on clear presumptions.

Speaker: Gunther Schober, Sales Manager at PSI Metals, Non Ferrous GmbH

Predicting the Future is a Matter of Trust

Gunther Schober, born November 21, 1970 in Leoben, Austria graduated in 1994 as Metallurgical Engineer with focus on industrial and energy economics at the Mining University of Leoben.

After his first experience as sales engineer and sales manager he took over management responsibility as head of sales & marketing for AVL DiTEST an international automotive company. 2007 he joined PSI Metals. As Senior Consultant, Project Manager and Sales Manager his expertise in metals and international experience in process development and Supply Chain Management is much sought-after.

With more than 10 years of experience in the area of Production Management Solutions comprising Supply Chain Management and Planning, Product Design, Quality Management, Production Execution and Logistics he became an acknowledged cognoscente in these fields. PSI Metals known as a very innovative company became metals industries’ partner for Industry 4.0.

Gunther Schober’s expertise together with the innovative approach of PSI metals make him a demanded contact for future solutions.



View the conference programme here

Sponsored by

Official Media Partner

Organised by

Supported by

Member of