摘要
在许多社会经济现象中，有这样的数学模型，它是不断地增长的，但又略有波浪式的起伏。要预测这类型的问题，既不能用一般的线性回归预测法，即使在ｎ＝２１时其相关系数高达０．９８８（甚至０．９９６），也能产生较大的预测误差。我们也不能用一般的曲线回归或多项式回归。包括很常用的龚珀资（Ｃｏｍｐｅｒｔｚ）曲线预测法也无济于事，因为龚法实际上是对具有Ｓ形数学模型的预测有效。它们都不符合波浪增长式且具有一定周期的数学模型。本文先就青岛市１９６２年至１９８２年这２１年人口资料进行分析探讨；再就１９７１年至１９９１年这２１年资料进行动态跟踪分析。并引进波浪增长曲线的作法。
In many social and economic phenomena .there is a mathematical model, which increases constantly and sometimes undulates a little. To forecast these kinds of problems, we can't use linear regression, it will make some forecast errors (see the example in the text). Also we can' t use the curve or polynomial regression; and the COMPERTZ curve is Y=Kabt, usually0<b<1, with s shape when t→∞, Y tends to K, it doesn't fit for the undulate increasing mathematical model with cyclic phenomena. This paper analyses and probes into the population data of a city from 1962 to 1982 A. D., and dynamic trace analyses are used for the same purpose with the data of the same city from 1971 to 1991 A. D. it leads to undulate increasing curve method.
出处
《数理统计与管理》
CSSCI
北大核心
1995年第1期7-10,共4页
Journal of Applied Statistics and Management
