[pt] Esta dissertação tem como objetivo comparar procedimentos de ajuste sazonal em séries temporais. As metodologias utilizadas são a do X12-ARIMA e a metodologia TRAMO/SEATS. Utilizaram-se as séries agregadas das Contas Trimestrais Brasileiras, fornecidas pelo Instituto Brasileiro de Geografia e Estatística - IBGE, no período compreendido entre o primeiro trimestre de 1991 e o segundo trimestre de 2000. Os aplicativos utilizados no decorrer do trabalho foram SPSS, FORECAST PRO, X12-ARIMA (versão DOS), SAS e TRAMO/SEATS (versão DOS). Também foram utilizadas séries simuladas com diferentes formulações para a tendência e sazonalidade, a fim de melhor analisar os resultados. / [en] This paper intends to perform a comparison of seasonal adjustment procedures. The compared methodologies are X12- ARIMA and TRAMO/SEATS. This current work encompasses the Brazilian Quarterly aggregated accounts which were obtained from the Brazilian Governmental Statistical Office (IBGE - Instituto Brasileiro de Geografia e Estatística) in the period between the first quarter of 1991 and the second quarter of 2000. This data, in the process of analysis, went trough the following software: SPSS, FORECAST PRO, X12-ARIMA (DOS version) and TRAMO/SEATS (DOS version). Some simulated series (with different structures for trend and seasonality) were also used in order to provide further and more accurate comparisons of the two methodologies. / [es] Esta disertación tiene como objetivo comparar procedimientos de ajuste estacional en series de tiempo. Las metodologías utilizadas son X12-ARIMA y TRAMO/SEATS. Se utilizaron las series agregadas de las Cuentas Trimestrales Brasileras, proporcionadas por el Instituto Brasilero de Geografía y Estadística - IBGE, en el período comprendido entre el primer trimestre de 1991 y el segundo trimestre de 2000. Los aplicativos utilizados en este trabajo fueron SPS, FORECAST PRO, X12-ARIMA (versión DOS), SAS y TRAMO/SEATS (versión DOS). También fueron utilizadas series simuladas con diferentes formulaciones para la tendencia y estacionalidad, a fin de analizar mejor los resultados.

Methods of analysis of seasonality in demografy Gabriela Myšáková Abstract The thesis presents statistical methods suited for analysis of seasonality in time-series. Three statistical methods have been thoroughly described, namely the time-series decomposition, the X12−ARIMA method and the cointegration of time-series. Further methods applicable for similar analysis have been briefly discussed as well. Three main methods have been subsequently applied to monthly demographic data for the Czech Republic and chosen European countries by natality, nuptiality and mortality. Seasonality has been discovered in all three demographic events and by using the time-series decomposition and the X12−ARIMA method for time-series lay out to separate units, and those progressions have been track by graphic and verbal interpretation. Cluster analysis has been applied to European countries nuptiality and mortality time-series in order to reveal similarities and dissimilarities among particular countries. All the methods were used onto data set by using appropriate procedures in statistical software SAS.