On the Automation of the Box-Jenkins Modeling Procedures: An Algorithm with an Empirical Test
提出一种量化博克斯-詹金斯方法中主观决策的自动算法，并通过经验数据验证其有效性，对时间序列建模研究者有参考价值。
In recent years, there has been an increased emphasis on the use of Box and Jenkins [1970] methods of modeling univariate stochastic time series.1 Unfortunately, the application of these methods involves a sequence of highly subjective decisions.2 The purpose of this note is to present an algorithm which quantifies this subjective process. Like any formal modeling process, the algorithm simply quantifies decisions and makes them automatically. In Section 1, I discuss the potential benefits of using an automatic univariate stochastic algorithm (USA). Section 2 presents the USA, followed by an empirical validation in Section 3. Section 4 contains the summary and conclusions.