Association function algorithm for decision tree
基于关联度函数的决策树分类算法

Variety bias is a prevalent problem existing in decision tree algorithms. For solving this problem, a new decision tree algorithm, AF algorithm, was proposed. The mechanism how it avoided this default was analysed. Experiments compared to ID3 algorithm show that AF algorithm can avoid the variety bias of ID3 algorithm, and has no worse performance in classifying instances then ID3 algorithm.