| dc.description.abstract | This master's thesis investigates the predictability of waiting times at crude oil ports
using Automatic Identification System (AIS) data and machine learning. Focusing on
the wet bulk market, specifically four congested Middle Eastern Gulf ports, we aimed to
answer: 11Can the waiting times in crude oil ports be predicted based on AIS data? 11. In
this thesis clustering algorithms with novel modifications are utilized to establish berth
and anchorage polygons. These polygons form the basis for a spatial matching of AIS
data that is used to generate event logs. A cross-sectional data set is derived from the
event logs which in turn is the basis for extracting features used in five different machine
learning models. The findings show that AIS-derived features have predictive power on
waiting times, with vessel composition within ports and port dynamics being significant
factors. These insights hold practical implications for ship owners and academics alike,
enhancing vessel economics through speed adjustments and facilitating further research
within the maritime domain. The thesis also proposes further research areas, including
methodology refinement within polygon generation, event log generation and waiting time
prediction. | en_US |
