基于粗糙熵的时序数据属性约简及规则提取研究

分析了近似质量在提取非确定性规则方面的不足，并基于粗糙熵的预测成功度概念，结合时序数据特点，提出一种属性约简及规则提取策略．该策略在对时序数据进行属性约简时，采用粗糙熵与时间距离相结合的方法，使得最终得到的约简在时序方面是较优的,最后使用UCI数据库进行仿真实验，效果良好．该策略在工程领域处理时序数据方面有一定的应用价值．

Research on the Attribute Reduction and Rule Acquisition of Time-Series Data Based on Rough Entropy

The deficiency of retrieving uncertainty rules with approximation quality was analyzed. According to the characteristics of time-series data,a strategy on data attribute reduction and rule acquisition using the concept of prediction success based on rough entropy was proposed.This strategy introduced a method that combined rough entropy and time distance to reduce the attributes of time-series data.The reduction obtained from time-series data had good time characteristics,Finally, simulation experiments using the UCI database confirmed that the algorithm had good effects.This strategy will be valuable in the field of engineering for the processing of time-series data.