针对隐含概念漂移和噪声的数据流,提出一种基于模糊积分融合的数据流分类方法(fuzzy integral ensembleclassifiers for mining data streams,FI-MDS)。将模糊积分融合方法与集成综合技术有效结合起来,首先通过基分类器对识别样例进行分类得到决策剖面,然后再用模糊积分融合方法得到最终的分类结果,同时引入动态权值更新以提高算法的适应性。实验结果表明,与传统的数据流分类算法相比,该方法提高了概念漂移的检测精度,有效地解决了数据流中复杂分类问题,具有良好的分类性和健壮性。
Compared with the extensive research on logistics network infrastructures(LNIs)in the developed world,empirical research is still scarce in China.In this paper the theory of LNIs is firstly overviewed.Then a new evaluation index system for LNIs is set up which contains factors that reflect the economic development level,transportation accessibility and turnover volume of freight traffc.An empirical study is carried out by using data envelopment analysis(DEA)and principal component analysis(PCA)approach to classify LNIs into 4 clusters for 25 cities in the Yangtze River Delta Region of China.According to the characteristics of the 4 clusters,suggestions are proposed for improving their LNIs.Finally,after comparing different LNIs of 25 cities in the Yangtze River Delta Region of China,this paper proposes that different LNIs including hub,central distribution center or cross docking center,regional distribution center or distribution center should be built reasonably in order to meet the customer's requirement in the four different cluster cities.
