Combining zero-inflated negative binomial regression with MLRT techniques: An approach to evaluating shipping accident casualties

Jinxian Weng, Dong Yang, Ting Qian, Zhi Huang

Research output: Journal article publicationJournal articleAcademic researchpeer-review

14 Citations (Scopus)

Abstract

This study aims to develop a maximum likelihood regression tree-based (MLRT) ZINB (zero-inflated negative binomial) model to predict shipping accident mortality, and also to examine the factors which affect the loss of human life in shipping accidents. Based upon 23,029 sets of shipping accidents observations collected from 2001 and 2011 in global water areas, a tree comprising 7 terminal nodes is built, each of which is assigned by a separate ZINB model. Model results indicate that the large number of shipping accident casualties are closely related to collision, fire/explosion, sinking, contact, grounding, operating time, capsizing, docking condition, hull/machinery damage, and miscellaneous causes. In addition, it is found that there is a larger casualty count for the accidents occurring under adverse weather conditions or far away from coastal/port areas. In addition, sinking is recognized as the accident type which causes the largest number of casualties. This study can help the decision makers to propose effective strategies to reduce shipping accident casualties.

Original languageEnglish
Pages (from-to)135-144
Number of pages10
JournalOcean Engineering
Volume166
DOIs
Publication statusPublished - 15 Oct 2018

Keywords

  • Maritime safety
  • Maximum likelihood regression tree
  • Negative binomial regression
  • Shipping accident

ASJC Scopus subject areas

  • Environmental Engineering
  • Ocean Engineering

Cite this