多维时空数据分布式协同进化属性约简分析

Attribute reduction analysis of distributed coevolution for multidimensional spatio-temporal data

面对多维时空数据属性复杂无法有效约简的问题,提出了多维时空数据分布式协同进化属性约简方法。通过多维时空数据投影抖动处理,使投影点分布角度一致。按照时间序列将多维属性的时空目标归入一维特征时空,构建约简布局优化结构。使用邻域粗糙集模型的贪心属性约简算法,获取多个属性约简子集,采用邻域覆盖约简规则学习方法,保证动态数据属性偏向约简过程稳定。根据协同进化约简布局优化结构,计算多维时空之间相对距离,获取偏序约简结果,实现多维时空数据分布式协同进化属性约简。由实验结果可知,该方法约简后数据最小为14个,具有良好约简效果。

A distributed coevolutionary attribute reduction method for multidimensional spatio-temporal data is proposed to solve the problem of complex attribute reduction for multidimensional spatiotemporal data.Through the projection dithering processing of multidimensional spatio-temporal data,the distribution angles of projection points are consistent.According to the time series,the spatial and temporal targets with multidimensional attributes are classified into one dimensional characteristic spatio-temporal,and the reduced layout optimization structure is constructed.The greedy attribute reduction algorithm of neighborhood rough set model is used to obtain several attribute reduction subsets,and the neighborhood coverage reduction rule learning method is used to ensure the stability of the dynamic data attribute bias reduction process.According to the coevolutionary reduction layout optimization structure,the relative distance between multidimensional spatio-temporal was calculated,and the partial order reduction results were obtained to realize the distributed coevolutionary attribute reduction of multidimensional spatiotemporal data.The experimental results show that the minimum number of reduced data is 14,and the method has good reduction effect.