First-Order Autoregression
First-Order Autoregression
复制标题
一阶自回归
DOI：
10.1007/978-1-935704-27-0_7
复制
发表时间：
1985
期刊：
影响因子：
--
通讯作者：
E. S. Epstein
中科院分区：
文献类型：
--
作者：
E. S. Epstein
在我们迄今为止所处理的所有情况中，我们都假定观察之间是独立的。然而，它在气象和气候时间序列中很常见（例如，月降水量，干旱指数），连续的元素是统计上相互相关的。持久性和周期性行为都是观察之间缺乏独立性的表现。在前一章考虑的例子中，尽管我们考虑了月与月之间的相关性，但从12月到下一个12月或从1月到下一个1月的随机元素的连续值都被明确地假定为独立的。换句话说，yidepended onxi但不是onyi-1。现在我们将考虑涉及时间序列的推断，其中每个成员的值在统计意义上取决于前一个值。
In all the situations with which we have dealt so far, we have assumed independence among the observations. However, it is commonplace in meteorological and climatological time series (e.g., monthly precipitation, drought indices) for successive elements to be statistically related to one another. Persistence and cyclical behavior are both manifestations of a lack of independence between observations. In the example considered in the previous chapter, the successive values of the random elementsϵifrom one December to the next or from one January to the next were explicitly assumed to be independent, even though we did allow for month-to-month correlation. In other words,yidepended onxibut not onyi−1. We will now consider inferences involving time series in which the value of each member depends, in a statistical sense, on the previous value.