《多表达式编程在规则挖掘中的应用研究》由会员分享，可在线阅读，更多相关《多表达式编程在规则挖掘中的应用研究（71页珍藏版）》请在金锄头文库上搜索。
1、 密级： 石家庄经济学院硕士学位论文 多表达式多表达式编程编程在规则挖掘中的应用研究在规则挖掘中的应用研究 论 文 作 者： 李阳 学 生 类 别： 全日制 一 级 学 科： 计算机科学与技术 学 科 专 业： 计算机应用技术 指 导 教 师： 张翠军 职 称： 教授 Secrecy Rate： Dissertation Submitted to Shijiazhuang University of Economics for The Master Degree of Computer Application Technology Study on the Application of Mul
2、ti Expression Programming in the Rule Mining by LiYang Supervisor: Prof.ZhangCuijun June 2014 原创性声明原创性声明 本人郑重声明：所呈交的学位论文，是本人在导师指导下，进行研究工作所取得的成果。除文中已经注明引用的内容外，本学位论文不包含任何他人或集体已经发表的作品内容，也不包含本人为获得其他学位而使用过的材料。对本论文所涉及的研究工作做出贡献的其他个人或集体，均已在文中以明确方式标明。本学位论文原创性声明的法律责任由本人承担。 学位论文作者签名： 日期： 关于学位论文版权使用授权的说明关于学位论文版
3、权使用授权的说明 本人完全了解石家庄经济学院关于收集、保存、使用学位论文的以下规定：学校有权采用影印、缩印、扫描、数字化或其它手段保存论文；学校有权提供本学位论文全文或者部分内容的阅览服务；学校有权将学位论文的全部或部分内容编入有关数据库进行检索、交流；学校有权向国家有关部门或者机构送交论文的复印件和电子版。 （保密的学位论文在解密后适用本授权说明） 学位论文作者签名： 日期： 导师签名： 日期： I 摘摘 要要 数据挖掘是一个多学科交叉融合的前沿学科，是信息技术发展过程中的重要成果之一，其理论研究和实际应用都已被广泛关注。关联规则挖掘是数据挖掘中的一项重要任务， 用于发现大量数据中项集之间隐
4、含的关联。 近年来， 随着互联网的迅速普及，数据库规模的不断扩大，关联规则挖掘变得越来越复杂，传统的关联规则挖掘算法已经不能满足当今知识发现的需要。学者们开始将多种算法应用到关联规则挖掘中来，其中演化算法是一种非常有效的方法，并已成为人们关注的焦点之一。 基因表达式编程是一种新的演化算法模型，主要借鉴了生物进化中基因表达的方式。多表达式编程是遗传程序设计领域的一个新的扩展。近几年来，多表达式编程因其独特灵活的编码方式，已经被广泛应用在诸多领域，并取得了一定的成果。 本文主要研究多表达式编程算法并将其应用于求解关联规则挖掘问题。首先，研究了基因表达式编程和多表达式编程的基本原理和实现技术；其次，
5、详细分析了数据挖掘中经典关联规则挖掘算法的优缺点；最后，针对关联规则挖掘问题，改进了多表达式编程算法，使其能更好的解决关联规则挖掘问题。 本文主要工作和创新点概括如下： (1)设计了多表达式编程算法的遗传算子和适应度函数。对数据挖掘中关联规则挖掘进行了深入研究、对传统关联规则挖掘算法进行了详细分析，结合关联规则挖掘问题、多表达式编程的编码方式、多表达式编程算法的优缺点，设计了适应度函数和几种不同的遗传算子。 (2)改进了多表达式编程算法。为了更充分的利用算法的优点，更好的弥补其不足，提出了改进的多基因表达式编程算法、改进的多表达式编程算法、改进的多表达式编程与差分演化融合算法、改进的多表达式编
6、程与模拟退火融合算法，并将这些算法应用于关联规则挖掘中。 (3)对算法进行仿真实验。通过对实验结果分析，证明了算法在求解规则挖掘问题时具有可行性，并表现出了很多优势。 关键词关键词：遗传程序设计；数据挖掘；多表达式编程；关联规则 II III ABSTRACT Data mining is a frontier subject of many subjects cross and fusion,it is one of the important achievements in the process of information technology development, nowaday
7、s its theoretical research and practical application have been concerned widely. Association rule mining is an important research task in data mining, it is used to find the implicit association among a large set of data items. In recent years, with the rapid popularization of the Internet and the c
8、ontinuous expansion of the size of the database, association rule mining is becoming more and more complex, the traditional association rule mining algorithms cannot meet the needs of data mining .Many scholars have begun to apply many kinds of algorithms to solve the problem of data mining, evoluti
9、onary computation is a very effective method and it has become one of the focus of attention. Gene expression programming is a new model of evolutionary computation, it mainly based on the way of gene expression in biological evolution.Multi Expression Programming is a new extension of the field of
10、Genetic Programming. In recent years, because of its unique and flexible encoding method, Multi Expression Programming has been widely applied in many fields, and also achieved certain results. This paper mainly studies the Multi Expression Programming algorithm and use it to solve the problem about
11、 association rule mining. Firstly, mainly studies basic principles and techniques of the Gene Expression Programming and Multi Expression Programming. Following detailly analyzes the advantages and disadvantages of the classical association rule mining algorithm. Finally, aiming at the problem of as
12、sociation rule mining, designs and improves of Multi Expression Programming algorithm, so that it can solve the problem of association rule mining more better. In this paper, the main work and innovations are summarized as follows: (1) The genetic operators and fitness function of Multi Expression P
13、rogramming algorithm have been designed.The association rule mining of data mining has been studied,the traditional association rule mining algorithm has been analyzed detailly. And combining with the question of association rule mining、the coding way of Multi IV Expression Programming and the advan
14、tages and disadvantages of Multi Expression Programming algorithm,fitness function and several different genetic operators. (2) The improvements about Multi Expression Programming algorithm. In order to make better use of the advantages,the better to make up for its shortcomings, put forward the imp
15、roved Multi Gene Expression Programming、 the combined algorithm of Multi Expression Programming and differential evolution algorithm、the combined algorithm of Multi Expression Programming and simulated annealing algorithm . And these algorithms are applied to solve the problem of association rule mi
16、ning. (3) The experiment of algorithms. Through the analysis of the experimental results prove the feasibility of the algorithm for solving the association rule mining problem.And results showed a lot of advantages. Keywords: Genetic Programming; Data Mining;Multi Expression Programming; Association RuleV 目目 录录 摘 要 .
