Meta-heuristic Methods for Outliers Detection in Multivariate Time Series

In this article we use meta-heuristic methods to detect additive outliers in multivariate time series. The implemented algorithms are: simulated annealing, threshold accepting and two different versions of genetic algorithm. All of them use the same objective function, the generalized AIC-like criterion, and in contrast with many of the existing methods, they don't require to specify a vector ARMA model for the data and are able to detect any number of potential outliers simultaneously. We used simulated time series and real data to evaluate and compare the performance of the proposed methods.