基于非随机初始种群遗传算法的分类规则挖掘

数据挖掘中分类问题一直是数据挖掘领域中研究的热点问题,先后提出了各种分类算法;其中遗传算法被认为是一种高效的分类算法.但是,传统的GA存在着易于陷入局部最优,致使得到的分类规则概括性不强的问题.提出了一种基于非随机初始种群的遗传算法分类规则挖掘算法.算法利用均匀种群方法生成非随机的初始种群,并通过均匀算子确保连续迭代过程中种群的多样性,从而达到防止GA早熟的目的.采用两个标准的公共领域的数据集验证了算法的有效性.实验结果表明,该算法能消除遗传算法在分类挖掘任务中收敛于局部最优的局限性,且能快速挖掘出易于理解的分类规则,提高对知识的理解力.

Mining Classification Rules Based on Genetic Algorithms with Non-random Initial Population

Classification in data mining is always the hot topic in data mining studies.Various kinds of classified algorithm are proposed successively among which genetic algorithm is considered as a highly effective classification algorithm.However,the traditional GA is easy to go and remain on stuck to a local solution,so the extracted rules have no agreeable generalization ability.Aimed at the question,a classification rule extraction method is proposed based on genetic algorithm with non-random initial population...