逐步回归自相关分析

作者从时间序列的自相关分析出发，提出了一种适于回归统计分析的生成预报因子的简单方法，建立起自相关预报模型。实例计算表明，此方法不仅预报精度较高，而且在对周期性相关因子的选择上具有等同的统计、检验前提，避免了一般方法因周期长度不同而带来统计，检验差异的影响。因而得到的回归因子较客观，真实。另外，还具有一定的分析时间序列隐含周期的能力。

Lag Correlation Analysis by Stepwise Regression Method

A new lag correlation analysis model is developed with a stepwise regression method. This model gives a simple method to generate predictors which suit to multivariate regression. Computational results show that this model is more objective in periodic predictor selection than general models. All predictors have the same premise of statistic and significance test to predictant.What's more,this model has some resolving power of hidden periodicity.