Biostats Club at PrevMed 11/12
Title: How many socks are in the laundry when the first recovered 11 are all unmatched?
– Also an example of Approximate Bayesian Computation using R
Presenter: Fridtjof Thomas, PhD, Department of Preventive Medicine
Wed., 11/12, 10am, Room 400 in the Doctors’ Office Building (66 N. Pauline, Memphis).
The likelihood function that expresses the probability of certain (observed) data under a particular statistical model can be costly to evaluate or even be unknown in complex real world situations. Approximate Bayesian Computation (ABC) bypasses the evaluation of a likelihood function and has for that reason gained some popularity in, e.g., population genetics, epidemiology and systems biology.
Our example for a non-standard problem deals with the situation where we would like to learn about the total number of socks we can expect in a laundry when we find only unmatched unique socks as the first 11 socks we pick. We formulate a sock picking simulation and demonstrate how ABC works (using the R statistical language) and which assumptions are critical in its execution. This example is taken from Rasmus Bååth’s research blog.