Abstract
Real world does not only provide noisy instead of perfect data. Every experimentalist has now and then to deal with outliers. The situation is simple if isolated points stick out of the general trend by a large amount. Arguments can then usually be found why such a point should be disregarded. The situation becomes critical if the outliers are not that obvious. This is usually the case for parameter space dimensions ≥ 3. We present a Bayesian solution to the outlier problem which assumes that the uncertainties assigned to the experimental data are only estimates of the true error variances.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
B. Cohen and B. Taylor Reviews of Modern Physics, 59, p. 1121, 1987.
P. Rousseeuw and A. Leroy, Robust Regression and Outlier Detection, John Wiley and Sons, New York, Chichester, Brisbane, Toronto, Singapore, 1987.
J. Taylor, An Introduction to Error Analysis, University Science Books, Sausalito, California, 1997.
D. Sivia, “Dealing with duff’ data,” in MAXENT96, Proceedings of the Maximum Entropy Conference 1996, M. Sears, V. Nedeljkovic, N. E. Pendock, and S. Sibisi, eds., p. 131, NMB Printers, Port Elizabeth, South Africa, 1996.
W. Press in Proceedings of Conference in Honor of John Bakcall, J. Ostriker, ed., p. in press, Princeton University Press, Princeton, 1998.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer Science+Business Media Dordrecht
About this paper
Cite this paper
Dose, V., Von Der Linden, W. (1999). Outlier Tolerant Parameter Estimation. In: von der Linden, W., Dose, V., Fischer, R., Preuss, R. (eds) Maximum Entropy and Bayesian Methods Garching, Germany 1998. Fundamental Theories of Physics, vol 105. Springer, Dordrecht. https://doi.org/10.1007/978-94-011-4710-1_4
Download citation
DOI: https://doi.org/10.1007/978-94-011-4710-1_4
Publisher Name: Springer, Dordrecht
Print ISBN: 978-94-010-5982-4
Online ISBN: 978-94-011-4710-1
eBook Packages: Springer Book Archive