Association rule learning to improve deficiency inspection in port state control

Abstract

The inspection of foreign ships in national ports is a critical measure in port state control (PSC), preventing substandard ships from entering national ports. Multifarious inspection items, limited inspection time and inspector manpower are challenging PSC inspection. This research applies data mining to analyze historical PSC inspection records in Taiwan’s major ports to extract potential valuable information for PSC onboard inspections. Using the Apriori Algorithm, the analysis identifies many useful association rules among PSC deficiencies in terms of specific ship characteristics, such as ship types, societies, and flags. The general rules identified show that the items ‘Water/Weathertight conditions’ and ‘Fire safety’ are significantly related. Besides, in the analysis of the various ship types, several different rules are found. After comparing the analysis of ship types and ship societies, it can be observed that the association rules for specific ship types, such as oil tankers, have a better effect than those for individual ship societies do. These identified rules can not only help inspectors effectively spot the associated deficiencies, but also improve the efficiency of PSC inspection. The ports other than Taiwan’s ports can apply a similar analysis method to identify corresponding association rules suitable for their own inspections.