基于粗糙集聚类算法的铁路通道客运市场细分

对铁路通道内客运市场的细分,是研究铁路通道客流分担、客运产品设计的基础. 根据宝鸡-兰州铁路通道旅客出行方式选择调查数据,结合粗糙集理论,首先,构建了铁路通道旅客出行方式选择决策表,对条件属性进行属性约简,并计算各属性的权重;其次,考虑到避免传统聚类算法的“维数陷阱”,提出了基于粗糙属性重要度的K-means聚类算法,并在UCI数据集上进行仿真实验;最后,运用该算法对调查数据样本进行聚类.结果表明：将铁路通道客运市场细分为6类时,具有最好的聚类效果;经统计分析发现,不同子市场的旅客出行行为有明显的偏好.

The Subdivision of Railway Passenger Transport Market Based on Rough Clustering Algorithm

The subdivision of railway passenger transport market is the basis for the study of passenger flow share rate in the railway corridor and the design of railway passenger transport products. According to the survey data on passengers travel mode choice in Baoji- Lanzhou corridor, combined with rough set theory, firstly, this paper builds the railway passenger travel mode choice decision table, makes attribute reduction on condition attributes, and calculates the weight of each attribute. Secondly, to avoid the "dimension trap" caused by the traditional clustering algorithm, it proposes the K-means clustering algorithm based on rough attribute significance, and makes simulation experiments on UCI data sets. Finally, using the clustering algorithm conducts the cluster on the sample of survey data. The results show that when the railway passenger transport market is divided into 6 classes, the clustering effect is the best, and the statistical analysis shows that the passenger travel behavior of different sub markets has obvious preference.