一种改进的数据挖掘算法

An Improved Data Mining Algorithm

针对Apriori算法的存在产生大量的候选频繁集合的缺点,本文提出了基于加权代价敏感的非频过滤矩阵Apriori算法,通过在FP-tree算法的基础上构造的决策树对应的数据进行代价敏感学习;设定不同的数据的权值,设定加权置信度;非频集过滤矩阵寻找频集,生成强关联规则;构成非频集过滤Apriori算法对应的初始矩阵;构建代价敏感的非频集过滤矩阵等措施提高了算法的挖掘效果。

Aiming at the Apriori algorithm＇s shortcoming of having a large amount of candidate frequent sets, this paper proposes the non-frequency filter matrix Apriori algorithm based on weighted cost and establishes corresponding data of decision based on the FP-tree algorithm to conduct the cost sensitive learning. Set different data weighted value and weighted confidence; non- frequency set filter matrix seeks the frequency set to generate strong association rules; compose the corresponding initial matrix of non-frequency filter Apriori algorithm; establish cost sensitive non-frequency filter matrix to improve the mining effect of the algorithm.