基于聚类和主成分分析的港口竞争力评价研究  

Port Competitiveness Evaluation Based on Cluster and Principal Component Analysis

作　　者：陈涵怡 金婷婷 尚雨浩 陈涵滢 

机构地区：[1]南京信息工程大学数学与统计学院,江苏 南京 [2]桂林理工大学信息科学与工程学院,广西 桂林

出　　处：《运筹与模糊学》2024年第1期947-959,共13页Operations Research and Fuzziology

摘　　要：随着全球化进程的加快,各国之间的贸易联系愈加紧密,港口已不再仅仅是联结水路运输、铁路运输与公路运输的枢纽,更是资金、技术、信息的中转站,是城市面向世界的重要窗口。本文采用主成分分析和聚类分析相结合的方法对我国年吞吐量在1000万吨以上的13个沿海港口的港口综合竞争力进行了分析和评价。我们首先构建了港口综合竞争力评价指标体系,包含4个一级指标和13个二级指标。然后我们通过主成分分析法从13个影响指标中提炼出三个主成分作为影响港口竞争力的新指标,并得出各港口三个主成分的得分。在这基础上,我们对各港口三个主成分得分求加权平均数,最终得出各港口综合竞争力的得分和排名。我们把根据主成分分析方法得出的港口综合竞争力排名与真实排名作对比,发现两者虽有出入但相差不大。根据各港口三个主成分的得分,我们进行层次聚类和K均值聚类,通过聚类个数与类内平方和、类间平方和的关系曲线图,选定适合的聚类数目。层次聚类和K均值聚类的结果在聚成两类时相差较大,但在聚成三类时则比较相似。With the acceleration of the process of globalization, the trade links between countries have become more and more close. Ports are no longer just the hub connecting water transport, railway transport and road transport, but also a transit station for capital, technology and information, which is an important window into the world. In this paper, principal component analysis and cluster analysis are used to analyze and evaluate 13 coastal ports with an annual throughput of more than 10 million tons. First, we study the factors that influence the port competitiveness, which can be divided into four aspects: The scale of port development, the conditions of port in-frastructure, the economic conditions in the hinterland of port and the development potential of port, from which we select 13 impact indicators, the evaluation index system of comprehensive competitiveness of ports is constructed. Then we use principal component analysis to extract three principal components from 13 impact indicators as a new indicator of port competitiveness, and get the scores of each port’s three principal components. On this basis, we weighted the sum of the three principal component scores of each port to get the overall competitiveness of each port score and ranking. Compare the ranking of port comprehensive competitiveness based on principal component analysis with the real ranking, and find that there are differences but not much difference between them. According to the scores of the three principal components of each port, we conduct hierarchical clustering and k-means clustering, and select the appropriate number of clusters by the graph of the number of clusters and the relationship be-tween the number of clusters and the sum of squares within and among clusters. The results of hierarchical clustering and K-means clustering are quite different when they are grouped into two groups, but similar when they are grouped into three groups.

关 键 词：主成分分析 层次聚类 港口综合竞争力 K均值聚类 

分 类 号：F55[经济管理—产业经济]