In this paper, a new algorithm to build incremental multivariate decision free is proposed. The advantages of the rough set theory and the multivariate decision nee are combined in this method. Aiming at the inconsistency between the new sample and the old sample, the core is computed. ff the core is empty, the generalization between core and decision attribute will be computed, the different results will be the different branches of decision tree at last. The decision tree rebuilding is avoided in the algorithm and the validity of the algorithm is proved by the example.%针对增量数据集,结合粗糙集理论和多变量决策树的优点,给出了增量式的多变量决策树构造算法.该算法针对新增样本与已有规则集产生矛盾,即条件属性相匹配,而决策属性不匹配的情况,计算条件属性相对于决策属性的核,如果核不为空,则计算核相对于决策属性的相对泛化,根据不同的结果形成不同的子集,最终形成不同的决策树分支.该算法很好地避免了在处理增量数据集时,不断重构决策树.实例证明该算法的正确性,对处理小增量数据集具有良好的性能.
展开▼
