基于基因表达式编程的人口预测模型

提出一种基于基因表达式编程算法(GEP)的人口预测新方法,并将该方法应用于东莞市人口预测实例问题研究。实验结果表明:由于基因表达式编程算法采用基因型与表现型相统一的编码方式、高效的遗传算子以及全局搜索的寻优方式,基于GEP算法的人口预测模型能够在样本少的情况下给出相对准确的预测结果。其验证数据的预测绝对值平均误差为0.96%,与灰色系统GM(1,1)预测模型及径向基人工神经网络预测模型相比,预测精度分别提高了18.34%、30.54%。GEP人口预测模型能够更好地挖掘人口发展的复杂非线性模式,有效防止过度拟合现象的发生,提供更为准确、合理的拟合及预测结果。

A Gene Expression Programming Algorithm for Population Prediction Problems

Predicting the size or development tendency of population is a complicated geographical problem. This kind of problem often involves multiple geographical components that interact in a complex way In this article, a new technique based on a gene expression programming (GEP) algorithm is presented, which can be used to address population prediction problems. In the context of GEP algorithm, population prediction problems are formulated by designing encoding strategies, evolutionary operations and fitness function. The population prediction model based on GEP approach is finally constructed and applied to predict population of Dongguan city. Compared with grey model and artificial neural network model, the predicting precision is improved by 18.34% and 30.54%, respectively. GEP model has better accurateness of predicting the size and development tendency of population. It can accurately fit nonlinear population development tendency and avoid overfitting to a certain extent. Gene expression programming algorithm can be used to effectively solve population prediction problems.