Output details

Meta-learning for time series forecasting and forecast combination

As part of collaborative research project (Lufthansa Systems, £100k, 2006-2009) this work investigated meta-learning for time series prediction. While originally developed in the context of demand forecasting for airlines, it later proved very versatile by winning the international transport-related time series forecasting competition held during WCCI 2010 conference (see related conference paper: 10.1109/FUZZY.2010.5584001).

The paper was listed in ScienceDirect's Top 25 Hottest Articles of the ‘Neurocomputing’ journal between April-December 2010 (http://top25.sciencedirect.com/subject/computer-science/7/journal/neurocomputing/09252312/archive/27/ - http://top25.sciencedirect.com/subject/computer-science/7/journal/neurocomputing/09252312/archive/30/). The meta-learning research is now further pursued in a funded PhD project (2013-2016) and contributing to the EU INFER project (€1,555,380, BU: €890,421, 2010-2014, Ref: FP7-PEOPLE-2009-IAPP-251617).