期刊名称：Scientific Annals of the “Alexandru Ioan Cuza” University of Iasi – Economic Sciences Series
印刷版ISSN：2501-1960
电子版ISSN：2501-3165
出版年度：2008
卷号：2008
出版社：Sciendo
摘要：At present, both at European Union and world level, experts are preoccupied to find the best method for the deseasonalisation of a time series that should assure the comparability of statistical data. The present paper follows the line of these researches.In the study, we undertake a comparison of the most representative methods based on moving average filter: moving average method, Census X-11 method and X-12 ARIMA method. Theoretical research shows the superiority of X-12 ARIMA method, which has incorporated the previous methods as regards the algorithm and the advantages, contributing to the improvement of the weaknesses of the former methods. The criteria for the comparison of the results obtained through the three methods applied to the time series of unemployment rate in Romania during the period 2000 - 2007 didn’t indicate a unique method, as being the most adequate for deseasonalisation.
关键词：seasonal variations ; moving average filter ; ARIMA models for stochastic processes
