基于聚类分析的铁路出行旅客类别划分

对铁路出行旅客进行类别划分，是简化旅客乘车选择问题研究的重要策略.根 据成都-武汉段既有线与新线的旅客调查数据，以旅客的各类主体、出行特性作为属性 变量，运用分层聚类法中的凝聚法进行变量聚类，将具有较大相关性的变量--时间价 值与月收入、出行目的与费用来源合并.然后根据简化的旅客属性变量指标集，运用近邻 传播算法对旅客进行样本聚类，并引用CH、Hart 及IGP 等聚类有效性指标确定最佳聚类 数.指标值表明，将铁路出行旅客划分为6 个类别时，具有最好的聚类效果.调查数据中旅 客乘车选择结果亦显示，不同类别的旅客对客运产品的选择有着明显的偏好.

Classification of Railway Passengers Based on Cluster Analysis

Classification of railway passenger is a crucial strategy of simplifying the problem of boarding choice for passengers. According to the survey data of Chengdu-Wuhan railway lines, this paper takes characteristic and travel features of passengers as property variables. Firstly, property variables are clustered by hierarchical clustering. The variables of great relevance such as time value and monthly income, trip purpose and cost sources are combined. Secondly, passenger samples are clustered by affinity propagation algorithms according to the simplified nodes indexes. Clustering effectiveness indexes contained CH, Hart and IGP indexes are analyzed to the clustering consequence. The result indicates that it is of the best effect while the passengers are divided into 6 sorts. The boarding choice of passengers in survey data also shows that different types of passengers give preferences to diversified transport product.