文件名称:Single-exponential-smoothing
介绍说明--下载内容均来自于网络,请自行研究使用
一次指数平滑法是指以最后的一个第一次指数平滑。如果为了使指数平滑值敏感地反映最新观察值的变化,应取较大阿尔法值,如果所求指数平滑值是用来代表该时间序列的长期趋势值,则应取较小阿尔法值。同时,对于市场预测来说,还应根据中长期趋势变动和季节性变动情况的不同而取不同的阿尔法值,一般来说,应按以下情况处理:1.如果观察值的长期趋势变动接近稳定的常数,应取居中α值(一般取0.6—0.4)使观察值在指数平滑中具有大小接近的权数;2.如果观察值呈现明显的季节性变动时,则宜取较大的α值(一般取0.6一0.9),使近期观察在指数平滑值中具有较大作用,从而使近期观察值能迅速反映在未来的预测值中;3.如果观察值的长期趋势变动较缓慢,则宜取较小的α值(一般取0.1—0.4),使远期观察值的特征也能反映在指数平滑值中。在确定预测值时,还应加以修正,在指数平滑值S,的基础上再加一个趋势值b,因而,原来指数平滑公式也应加一个b。-Single exponential smoothing refers to a first end of exponential smoothing. If in order to make the exponential smoothing value is sensitive to reflect changes in the latest observations, should take a larger Alfa value, if the exponential smoothing value is used to the long-term trend represents the values of the time series, should take the smaller Alfa value. At the same time, the market forecast, also should take different Alfa value, according to changes in long-term trend and seasonal changes of different in general, shall be treated according to the following situation: 1 if the long-term trend observed values close to constant changes, should take the middle value (generally 0.6- 0.4) to observe value has a size close to the weight in the index smoothing 2 if the observed values showed clear seasonal changes, it is desirable to take larger values (usually 0.6- 0.9), the recent observation has important role in exponential smoothing value, thereby enabling the observation valu
(系统自动生成,下载前可以参看下载内容)
下载文件列表
Single exponential smoothing.txt