摘要： 生成用于预测的关联规则，现有算法生成的关联规则中许多是不必要的。利用分治策略和基于频繁闭子集的FP-TREE 生成一种特殊的关联规则集(最小预测集)，它比现有规则集小，但是具有同样的预测功能，并且具有发现关键属性的能力。给出了它的算法，并从理论上证明了该算法的正确有效性。

关键词: 数据挖掘；关联规则；频繁闭项集；最小预测集；FP-tree

Abstract: Mining transaction database for association rules usually generates a large number of rules ,most of which are unnecessary when used for subsequent prediction and discovery of key data. In this paper, a particular set called the minimal prediction rule set, is mined using the divide-and-conquer strategy and the FP-tree of frequent closet item sets. It is smaller than the association rule set, but has the same ability of prediction and discovery of key data. In this paper, the algorithm of the minimal prediction rule set is given, and its effectiveness is proved in theory.

Key words: Data mining; Association rules; Frequent closet item sets; Minimal prediction rule set; FP-tree