In addition to numerous interesting presentations given by scientists and practitioners from all over the world, the conference offered plenty of opportunities for exchanging ideas and networking at various evening events.
We also had the opportunity to present some of our current research projects in the field of forecast combination in the stream Analytics, Data Science, and Forecasting:
Veronika Wachslander presented a method for determining forecast combination weights under the title “Shifting Forecast Combination Weights Estimated on Subsamples towards Full Sample Weight Determination”. This method uses resampling of training data to repeatedly estimate combination weights and subsequently shifts these weight vectors towards more stable weighting schemes. The final combination weights are obtained by aggregating all weight vectors.
André Konersmann presented the results of a joint research project with Thomas Setzer titled “Generating Artificial Training Errors for Forecast Combination Based on Error Vector Permutation”. The method aims to stabilize the sample covariance matrix by generating new synthetic error instances based on permuted forecast error vectors. The augmented data set is then used to calculate combination weights.
We would like to thank the GOR and Bielefeld University for the excellent organization of this informative conference!