Machine Learning for Classifying Certificate of Competency Applications

Abstract

In this thesis we research the possibility of classifying Certificates of Competency. An application for such a certificate consists of selecting which competencies you want, providing various seagoing claims and providing various diplomas. We try to create a machine learning model that, based on all these knowledge, is able to correctly classify whether the application should be accepted or denied. Our main conclusion is that it is possible, as long as the labels are updated. There are multiple binary classifiers that can be used but only the boosted decision tree worked consistently on each problem we tested.