Scan Statistics: Theory and Applications

Probabilités et Statistique

Salle séminaire M3-324
Alexandru Amarioarei
Université Lille 1
Mercredi, 5 Mars, 2014 - 10:30 - 11:30

The object of this talk is to present some of the work done in the eld of discrete
scan statistics. After introducing the framework and the problem, we provide a
series of approximations and bounds for the probability distribution of one, two
and three dimensional scan statistics over a random eld generated by i.i.d. obser-
vations. We also present an importance sampling method for the simulation and
we compare it with the direct Monte Carlo approach. We nish by considering the
problem of discrete scan statistic over a random eld generated by a block-factor
type model obtained from i.i.d. sequence (for example over Moving Averages of
order q). Numerical results for di erent scenarios will be provided (some of them
live, using a Matlab GUI).