On Tuesday 29 September 2020, the Avon local group held an online meeting on 'Data Science at Electricity Providers'. The first speaker was Jethro Browell, a lecturer and EPSRC Innovation Fellow at the University of Strathclyde where he leads the Energy Forecasting Group. Jethro's talk, titled 'Challenges and Best practice in Energy Forecast Evaluation', gave an overview of the role of forecasting in the electricity industry and then focussed on the importance of choosing performance metrics that are relevant to the specific application of interest.
The second speaker was Yannig Goude, a senior researcher at Électricité de France (EDF) R&D Lab since 2008 and an associate professor at the Universitè Paris-Saclay. In his talk, titled 'Adaptive Methods for Short-Term Electricity Load Forecasting During COVID-19 Lockdown in France', Yannig described the effect of the COVID-19 pandemic and subsequent lockdown on the aggregate electricity demand in France, and described a novel forecasting approach that is able to adapt to abrupt changes in demand patterns.
The third speaker was Stefan Birr, who work as Data Scientist at E.ON Digital Technology since 2017. Stephan's talk, titled 'Forecasting Seasonal Time Series', discussed the challenges of forecasting customer contacts (eg, calls, letters and emails from customers) and assessed the predictive performance of several machine learning and statistical forecasters on this application.
The last speaker was Ria Van Hecke, who works as Data Scientist at E.ON since 2018. Ria's talk was titled 'Knowledge Management for Energy Sale' and focussed on Soφia, a complex knowledge management tool developed at E.ON. Ria explained how Soφia was born as a knowledge generator for the B2B strategy team at E.ON, embedding complex tools for news topic modelling and classification, and then describe how it has been extended to provide customer leads recommendations and to produce predictions on future tech trends.
The workshop was very well attended, reaching around one hundred virtual participants, and the attendee's questions between each talk lead to stimulating discussions.
Written by Matteo Fasiolo, a Lecturer in Statistical Science in the School of Mathematics at the University of Bristol.
Watch a recording of the meeting: