Drift Mining is a technique to analyse the evolution of data distributions in order to understand and predict their drift (change over time).
Publications
- When Learning Indeed Changes the World: Diagnosing Prediction-Induced Drift BibTeX PDF (author's manuscript)
-
Georg Krempl, David Bodnar, Anita Hrubos
In: T. De Bie and E. Fromont (eds.) Advances in Intelligent Data Analysis XIV - 14th Int. Symposium, IDA 2015, St. Etienne, France. Published by Springer, to appear at http://www.springer.com/series/7409.
- Temporal Density Extrapolation BibTeX PDF (author's manuscript)
-
Georg Krempl
In: A. Douzal-Chouakria, J. Vilar P.-F. Marteau, A. Maharaj, A. Alonso, and E. Otranto (eds.) Proceedings of the ECML/PKDD 2015 Workshop on Advanced Analytics and Learning on Temporal Data (AALTD 2015). Porto, Portugal. September 11, 2015 Published by CEUR Workshop Proceedings, to appear at http://SunSITE.Informatik.RWTH-Aachen.DE/Publications/CEUR-WS.
- Drift mining in data: A framework for addressing drift in classification BibTeX PDF (author's manuscript)
-
Vera Hofer, Georg Krempl
Computational Statistics and Data Analysis 57(1) 2013, p. 377-391. Published by Elsevier, available at http://www.sciencedirect.com/science/article/pii/S0167947312002812.
Implementations
(coming soon)
People
Methods for drift mining are developed by a team of researchers at the KMD lab, Otto-von-Guericke University (OvGU) Magdeburg, Germany, and the Dep. of Statistics and Operations Research, Karl-Franzens University Graz, Austria.
Contributors include:- Georg Krempl, KMD Lab, Otto-v.-Guericke University Magdeburg, Germany
- Vera Hofer, Statistics and Operations Research, Karl-Franzens-University Graz, Austria
- Myra Spiliopoulou, Head of KMD Lab, Germany