Large Scale Predictive Analytics – From Lab to Fab
Dr. Martin Hahmann
Dresden University of Technology (TUD)
The manufacturing industry has openly embraced the introduction of Big Data Analytics into its processes. Especially predictive analytics benefit manufacturers, as they offer valuable insights regarding tool capabilities, line congestion, maintenance, and much more. With pilot projects already in action, the future will see predictive analytics on a much bigger scale. Predictive models will become ubiquitous in factories, forming large model landscapes that will take in data from a multitude of sources.
This scale-up brings new challenges, of which we address some in this talk. To fully utilize the available data, predictive models must become fast, robust and easily adaptable. As available data sources constantly increase, benchmarks, automatic model selection and efficient model maintenance become necessities for sustaining an optimal and up-to-date model landscape. Besides these pure technical aspects, the human in the loop also must be considered. Analytic systems still need to work together with people, who determine the goals for the system, provide part of its input and need to work effectively with the output.
- 2007 diploma in computer science from TU Dresden
- since 2007 researcher at the Database Technology group, participation in different research projects including collaborations with SAP, GfK, Global Foundries, Systema…
- 2013 PhD from TU Dresden with a dissertation on feedback-driven data clustering
- since 2014 member of the german competence center for scalable data services and solutions (ScaDS)