Genetic Algorithms For Outlier Identification In Time Series

With the development of modern information technology, the importance of outliers ,which may contain a lot of useful information, has been being paid more attention to as people handle with data in time series. So, more and more scholars have committed themselves to identifying outliers .In this paper,a genetic algorithm is proposed for identifying additive and innovational outliers in ARMA and ARMAX series.The genetic algorithm allows simultaneous comparison of a great number of outliers' patterns. This way we may hope to overcome the problems of masking and swamping,that are very likely to arise in the comtext of time series analysis with aberrant observations.We are using the standard genetic algorithm with complete replacement of the past population and elitist strategy.lt is apparent that fast computation is necessary for the genetic algorithm approach to turn out viable in practice. The relationship between inverse correlations and outliers is helpful to simplify the fitness function,which may be quickly computed by Trench's algorithm.For the case of large series,it is better to devide the series into several parts so that the two neighbour subseries share a length of same data,and then to detect the subseries respec-tively.Thus,the problem of large population in genetic algorithms has been avoided ,as well as the loss of outliers around the cut point.Some case studies show that the algorithm is effective in detecting outliers' location and type and in estimating their size.