基于动态约简的贝叶斯分类器实现

海量数据且高维环境下,朴素贝叶斯分类可能即面临获取大量带类标签代价过高又面临当前分类规则不能适应数据变化等问题。于是提出一种基于小规模训练集的基于粗糙集(RS)动态约简贝叶斯算法来实现问题分类:利用粗糙集理论对决策表属性进行动态约简,挖掘出对分类最有利的条件属性即极小值属性,作为朴素贝叶斯推理(NBC)方法对知识进行学习和分类的输入。该方法结合了贝叶斯推理与动态约简将大数据库采样划分的优点。实验证明了算法的可行性。

Rough set-based Dynamic Reduct Bayesian classifier to achieve

In case of huge data sets and the data with higher dimentional space,Naive Bayesian may face the problems that get most class labels with costly and that the current classifying rules can't apadt the varied data.So,an incremental Naive Bayes algorithm based in rough-set dynamic Reduct has been brought up.The algorithm is shown as following:attributes of conditional attributions has been dynamic reduced based on rough set to get the minimum attributes.Then naive Bayesian method with reduced attributions has ...