模糊划分的决策树方法

Decision Tree Based on Fuzzy Discretization

在许多优化问题中，目标值是连续的。对这类问题，首先对目标值进行离散化，再采用决策树方法提取规则。在一定程度上，相比直接对连续的目标值优化可提高正确率，并增加结果的可理解性。为了克服分段划分带来的突变性，可将目标值进行模糊划分，再采用决策树方法提取规则，这样进一步可提高正确率。

The targets of many optimization problems contain continuous value. In order to solve these problems conveniently, discretization is applied on them as a preprocessing step. Compared with the algorithms without such step, this usually improves the accuracy of the result and enhances its understandability. However ordinary discretization usually results in drastic and unreasonable change in its result. To overcome this Problem this paper proposes a new method that discretizes the continuous target using fuzzy logic, and then uses decision be to extract the rules lies. in the training data. Experiments involved in this paper prove that this method could improve the accuracy further more.