摘要
基因表达式编程(GEP)是将进化的遗传操作和个体的适应度评价相分离的进化模型,具有比遗传编程快2-4个数量级能力。邮电业务量是反映经济发展的重要指标之一,其预测技术得到了广泛的研究,主要工作包括:阐述了GEP基本原理,以及GEP进行时间序列分析的基本方法;运用了GEP技术,对邮电业务总量进行建模研究,并进行了预测检验和分析。实验结果表明,基于GEP得到的邮电业务总量模型有较好的泛化能力,在测试数据上平均相对误差为4.44%。
Gene expression programming （GEP） is a novel evolution system. It is featured with separation of genetic operation and fitness evaluation. GEP is faster 100-60000 times than its predecessor GP. Business total of post and telecommunication is one of important factors which reflect economic developments, so its forecasting techniques are developed and studied all the time. Firstly, the principle of GEP, and the basic methods applied GEP to time series analysis are introduced. Secondly, based on GEP technology, business total of posts and telecommunication is modeled, and this model is tested and analyzed. Finally, illustrate experimental result show this model has a good generalization ability, which has mean relative error 4.44% on the test data.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第18期4124-4127,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(60773169)
江苏技术师范学院博士启动基金项目(KYY09001)
关键词
基因表达式编程
邮电业务量
进化模型
预测
泛化能力
gene expression programming
business total of posts and telecommunication
evolutionary model
prediction
generalization ability
