【摘要】 该文根据决策树结点的错误率与分割后的错误率减少量,提出一种新的基于遗传程序设计(GP)的GP-决策树优化算法。该算法不但可以求解出GP-决策树结点的权值矢量,同时也确定了GP-决策树的结构。实验结果表明,应用GP-决策树优化算法能够正确完成对趋势预测模型的选择。更多还原

【Abstract】 In this paper we introduce a new GP-decision tree optimization algorithm based on genetic programming,according to decrement of tree node error ratio and divided error ratio.This algorithm can not only solve the weights’vector of GP-decision tree node,but also ascertain the construct of two-member decision tree.Experimental results show the choice for trend forecasting models can be correctly finished by using GP-decision tree algorithm.更多还原