关于自回归预测方法的注记

A note on the method of autoregressive prediction

已知时间序列资料,建立自回归模型进行预测时,为了提高预测的准确性,采取的途径之一是固定自回归模型为线性模型,对数据列进行变换,提高数据列的光滑程度.证明了利用反双曲正弦函数变换能提高数据列的光滑程度,给出了改善的自回归预测方法,并且举例加以论证。

When constructing an autoregressive model to predict according to time series data, in order to improve the veracity, one of the methods is to let the model he a linear one, transform the data row and increase the smooth degree of them. This paper show that the transformation using are-hyperbolic sine function can smooth the data row. Then an improving autoregressive method is proposed and some examples are given to verify it.