基于非线性相关发现的数据挖掘算法

Data Mining Algorithm Based on Non-linear Correlation Discovery

现有的关联规则挖掘算法均致力于频繁集搜索,基于预先设置的支持度—置信度之上,具有很大的偶然性,不利于控制;并且关联规则没有体现数据整体的相关性。为了克服以上缺点,引入了非线性相关的概念,应用于不同相关类型规则的挖掘,且无须人为设置参数,从而大大提高了规则发现的实效性。

The existing arithmetic of association rules mining techniques all do take up with searching the frequent setting, especially on the basis of support-believe degree set beforehand. So the searching has much more chance going against control. In addition, the association rules are out of correlatino of holistic data. Importing the concept of non-linear correlation, applying to different kinds of rules mining, which conquered those shortcoming discussed above. The proposed method need not set parameter man-made, and it enhances the substantial results of finding the rules greatly.