Model selection and statistical inference
May 3 – 5, 2011, LAUSANNE
Jointly organized by the Doctoral Program in Ecology and Evolution
and the Doctoral Program in Population Genomics
Prof. Nigel Gilles Yoccoz, University of Tromsø, Norway
The goal with the course is to provide an overview of different approaches (including how they can be implemented), and not to advocate a specific approach.
Simulations as well as real data sets will be used to illustrate the different approaches and why they might give different answers.
Box wrote that “all models are wrong but some are more useful than others” (Box GEP, Hunter JS, Hunter WG. 2005. Statistics for experimenters. Design, innovation, and discovery. Hoboken, New Jersey: Wiley-Interscience, John Wiley & Sons), and the same can be said about model selection approaches – they are all wrong but some work better than others!
Some suggested reading is provided below, trying to cover both the ecological/evolutionary literature and the statistical literature (so as to provide a diversity of points of view). More references/papers will be provided before and during the course.
Day 1: Concepts and theory
What is a statistical model and what it is used for?
Cox DR. 1990. Role of models in statistical analysis. Statistical Science 5: 169-174.
How do we measure statistical evidence – P-values, Likelihood, AIC, Bayes factor.
Different chapters in Taper ML, Lele SR, eds. 2004. The nature of scientific evidence Chicago: University of Chicago Press.
Assessing the goodness of fit of a statistical model
Any good statistical textbook! A very thorough is Cook RD, Weisberg S. 1999. Applied regression including computing and graphics. New York: John Wiley & Sons.
Day 2: Implementations of different approaches to model selection
Models as predictive tools: how do we measure predictive ability and how do we estimate it?
Hastie T, Tibshirani R, Friedman J. 2009. The elements of statistical learning. Data mining, inference and prediction. New-York: Springer-Verlag.
Models as tools to estimate parameters
Claeskens G, Hjort NL. 2008. Model selection and model averaging. Cambridge: Cambridge University Press.
Day 3: Uncertainty and model averaging
What is uncertainty and how do we represent it?
Regan HM, Colyvan M, Burgman MA. 2002. A taxonomy and treatment of uncertainty for ecology and conservation biology. Ecological Applications 12: 618-628.
The consequences of ignoring model selection for prediction/parameter uncertainty
Chatfield C. 1996. Model uncertainty and forecast accuracy. Journal of Forecasting 15: 495-508.
Burnham K, Anderson D, Huyvaert K. 2011. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behavioral Ecology and Sociobiology 65: 23-35.
It is strongly recommended you refresh your basic probability theory (in particular conditional probability, Bayes theorem) and statistics! R will be used throughout; a list of required libraries will be provided before the course.
Date: May 3-5, 2011; 3 day course
Time: 10 am - 6 pm
Location: University of Lausanne, Biophore building, Seminar room 2213
Number of participants: max. 25
Participants need to bring their own laptop.
Please send an Email to the coordinator, if you would like to be added to the waiting list or to state your interest for a similar course in the future.
This is a joint course between the Doctoral Program in Ecology and Evolution
and the Doctoral Program in Population Genomics.
Therefore, priority is given to PhD students of these doctoral programs.
Coordinator of the Doctoral Program in Population Genomics
Le Biophore, UNIL-Sorge
University of Lausanne