Jan de Gooijer – ASE
Detecting Change-points in Multidimensional Stochastic Processes
Abstract of the presentation:
Procedures for detecting changes of variances in an ordered sequence of (possibly independent) observations taken from a multidimensional stochastic process can
help to elucidate the structure of the process. For instance, it is well known that homogeneity of variance in a sequence of observations taken from a single financial risk factor does not necessarily imply a homogeneous behaviour in variance of all possible risk factors simultaneously. Hence, a univariate change-point detection procedure may well fail to reject the assumption of constant variance underlying the model fitted to each individual sequence of observations.
A general test statistic for detecting change-points in multidimensional stochastic processes with unknown parameters is proposed. The test statistic is specialised to the case of detecting changes in sequences of covariance matrices. The finite-sample properties of the test statistic are compared with two other test statistics. Using a binary segmentation procedure, the potential of the various test statistics is investigated in a multi-dimensional setting both via simulations and the analysis of a real life example.
Lunch is provided during the presentation.
More information: Cesar Ariza: C.J.ArizaRojas@uva.nl
