SPEAKER: |
|
TITLE: |
"Mixture distributions" |
DAY: |
Wednesday, January 20, 1999 |
TIME: |
3:30 p.m. [Coffee in BSB-202 at 3:00 p.m.] |
PLACE: |
BSB-108 |
Fitting mixture distributions to samples from heterogeneous populations raises some interesting statistical and computational issues. I will outline a likelihood-based approach that works well in practice, and illustrate the methodology with a variety of examples from biology, medicine, finance, geology and archeology. We will see that a model which fits well isn't necessarily correct, and a model that fits poorly isn't necessarily useless. The presentation will rely heavily on graphics.
Peter Macdonald received his Bachelor's and Master's degrees in Mathematics from the University of Toronto, and his D.Phil. in Biomathematics from the University of Oxford. He joined McMaster in 1971 and has spent research leaves at l'Institut National de la Santé at de la Recherche Médicale in Villejuif, France, and La Trobe University, Bundoora, Australia. He has held many positions in the Statistical Society of Canada, including that of President in 1990-91. His MIX software for fitting mixture distributions is used world-wide.
A good introduction to mixture distributions can be found at the MIX web site. Several of these examples will be discussed in the seminar.
The following references have been provided by Dr Macdonald to be used as background for his talk. They are on reserve at Thode Library (STATS 770: Statistics Seminar).
[1] McLachlan, G.J. and K.E. Basford (1988). Mixture Models: Inference and Applications to Clustering. Marcel Dekker, New York. xi+253 pp.
[2] Pearson, K. (1894), "Contributions to the mathematical theory of evolution," Phil. Trans. Roy. Soc. A, 185, pp. 71-110.
[3] Titterington, D.M., A.F.M. Smith and U.E. Makov (1985). Statistical Analysis of Finite Mixture Distributions. Wiley, New York. x+243 pp