Calendar Effects in Monthly Time Series: Detection by Spectrum Analysis and Graphical Methods
本文介绍了两种检测月度时间序列中日历效应的方法：频谱分析和时域图形展示，帮助分析师判断是否需要调整日历效应，并验证调整效果。
Abstract Many time series, particularly national business and economic series, which are reported monthly and represent a total of the series for each month, contain calendar effects due to changing month length, weekly periodicities, and holidays. It is important to detect and remove this spurious calendar variation to allow a better appreciation of the variation in the series due to important factors. This article discusses detection. Two sets of diagnostic methods for detecting calendar effects in monthly time series, spectrum analyses and time domain graphical displays, are described. These methods can be used in an initial analysis to decide if calendar adjustment is necessary and can be used on an adjusted series to determine if the adjustment has properly removed all of the calendar effects.