Marine accidents analysis based on data mining using K-medoids clustering and improved A priori algorithm

2018 International Conference on Civil and Hydraulic Engineering

土木建筑工程;水利工程

Yang, Baicheng^1 ; Zhao, Zhilei^1 ; Ma, Jianbin^2

Navigation College, Dalian Maritime University, Dalian, Liaoning

116026, China^1

Shenzhen Urban Public Safety and Technology Institute, Shenzhen, Guangdong

510852, China^2

关键词: Classification mining;    Collision accidents;    Confidence threshold;    K-medoids clustering;    Maritime accidents;    Mining algorithms;    Negative association rules;    Support threshold;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/189/4/042006/pdf
DOI  :  10.1088/1755-1315/189/4/042006

学科分类：土木及结构工程学

来源: IOP

【 摘 要 】

In order to analyze the causes of maritime accidents and further ensure the safety of ship navigation, the risks of ships at sea and their impact factors are identified. The Zhejiang Sea area, China is used as the case study. Real maritime accident data of this area are collected and used as model inputs. The accident types are clustered center, and merges with the A priori algorithm to realize the classification mining of maritime accident data. 8 association rules of collision accidents were extracted under the conditions of 20% of the support threshold and 50% of the confidence threshold after removing negative association rules. The mining algorithm is based on K-medoids and A priori combination. K-medoids improves accuracy of results. In this paper, lift value is introduced to assist the evaluation of the association among these factors, which enhances the stability of results, compared with traditional methods. Finally, by analyzing the association rules one by one, the causes and characteristics of maritime accidents in Zhejiang waters are summarized.