时间序列分析中指数平滑法的应用

Application of Exponential Smoothing in Time Series Analysis

对于平稳时间序列分析,最常采用4种方法即指数平滑法,自回归法,自回归综合移动平均法和季节分解法。本文通过具体实例分别介绍指数平滑中三种不同方法即Simple法,Holt法和Winters法的应用。它们是分别适用于时间序列数据集无趋势和季节变化,有线性趋势无季节变化和有季节变化的模型分析法。

For stationary time series analysis,there are four normal methods,namely exponential smoothing method,a-utoregression method,ARIMA method and seasonal decomposition method.Through examples,this paper introduces three kinds of exponential smoothing,namely Simple method,Holt method and Winters method.They are respectively applicable to the time series data of no trend and no seasonality,of linear trend and no seasonality and of linear trend and multiplicative seasonality.