从熵均值决策到样本分布决策

为了研究归纳学习的判决精度问题,分析了C4.5算法的不足以及标准算法与亚算法之间争论和妥协的根本原因,从估计训练样本的概率分布的角度出发,给出了一种简单而新颖的决策树算法.基于UCI数据的实验结果表明,与C4.5算法相比,该方法不仅具有比较好的判决精度,而且具有更快的计算速度.

Decision Varied from Entropy to Parametric Distribution

In order to improve the predictive accuracy of inductive learning, a heavy analysis about the demerit of C4.5 is given, and the reason why there are many debates and compromise between standard method and meta algorithms is pointed out. By the method of estimating the probability distribution of training examples, a new and simple method of decision tree is turned out. Experimental results on UCI data sets show that the proposed method has good performance on accuracy issue and faster computing speed than C4.5 algorithm.