Module title:<\/strong> Advanced Maritime Safety Analytics and Human Factors<\/p>\n Assessment type:<\/strong> Individual technical FSA-style written assignment<\/p>\n Weighting:<\/strong> 40 percent of module grade<\/p>\n Length:<\/strong> 3,000\u20134,000 words (excluding reference list, tables, figures and appendices)<\/p>\n Submission format:<\/strong> Word processed report (DOCX or PDF) via the VLE or learning portal<\/p>\n Level:<\/strong> Final year undergraduate or postgraduate taught (Level 6 or 7 equivalent)<\/p>\n Maritime administrations, coastal states and operators face increasing pressure to improve safety performance while simultaneously decarbonising, digitalising and optimising operations. The growing availability of AIS data, casualty databases, weather archives and operational datasets has made machine learning a practical tool for predicting accident risk, supporting inspection targeting and informing navigational safety policy. At the same time, accident investigation findings continue to demonstrate that human performance, organisational culture and decision making remain central contributors to incidents and near misses, even in highly automated environments. This assessment requires the development of a structured, FSA-style analysis that critically examines how machine learning based risk prediction can be integrated with human factors perspectives to support safer and more sustainable maritime operations.<\/p>\n Prepare a 3,000\u20134,000 word technical FSA-style report<\/strong> evaluating the use of machine learning for maritime accident risk prediction, with explicit consideration of human and organisational factors in sustainable ship or port operations.<\/p>\n You may focus on one of the following areas, or agree an equivalent topic with your tutor:<\/p>\n Machine learning based prediction of collision and grounding risk in a defined coastal or port approach<\/p>\n<\/li>\n Machine learning supported targeting of port state control or flag inspections for safety and environmental risk reduction<\/p>\n<\/li>\n Machine learning enabled decision support for vessel traffic services or traffic management in congested routes<\/p>\n<\/li>\n Machine learning driven risk forecasting for a specific vessel segment such as tankers, ferries, Ro-Ro vessels or offshore support vessels<\/p>\n<\/li>\n<\/ul>\n Your report should be framed in accordance with the steps of the IMO Formal Safety Assessment, adapted to the analytical rather than regulatory purpose of this assignment.<\/p>\n Structure your analysis using clear headings and subheadings.<\/p>\n i. Problem definition and system description<\/strong><\/p>\n Define the maritime system or route under study, including vessel types, traffic characteristics, environmental conditions and operational context<\/p>\n<\/li>\n Clearly explain the safety problem addressed, such as collision hotspots, high risk vessels, pilotage incidents or near miss patterns<\/p>\n<\/li>\n<\/ul>\n ii. Hazard identification and risk model<\/strong><\/p>\n Identify key accident types and contributing factors across technical, environmental, human and organisational domains<\/p>\n<\/li>\n Present a conceptual risk model linking predictors such as vessel characteristics, traffic density, weather conditions and human factor proxies to accident outcomes<\/p>\n<\/li>\n<\/ul>\n iii. Machine learning approach for risk prediction<\/strong><\/p>\n Describe a suitable machine learning approach, such as logistic regression, random forests, gradient boosting, AutoML or neural networks, drawing on recent maritime studies<\/p>\n<\/li>\n Discuss data sources, feature selection, model training, validation and performance metrics, demonstrating understanding of the workflow without requiring implementation<\/p>\n<\/li>\n Explain how predicted risk scores or probabilities could be applied in practice, for example for inspection targeting, routing advice or traffic control<\/p>\n<\/li>\n<\/ul>\n iv. Integration of human and organisational factors<\/strong><\/p>\n Analyse how human factors including workload, situation awareness, bridge resource management and safety culture shape both accident risk and the use of machine learning tools<\/p>\n<\/li>\n Consider potential new failure modes such as automation bias, over-reliance on algorithmic outputs, data misinterpretation or erosion of professional judgement<\/p>\n<\/li>\n<\/ul>\n v. Risk control options and cost benefit perspective<\/strong><\/p>\n Identify and compare risk control options that combine machine learning based prediction with procedural, training and organisational measures<\/p>\n<\/li>\n Provide a qualitative or semi quantitative cost benefit discussion consistent with FSA thinking, addressing effectiveness, feasibility, implementation effort and sustainability impacts<\/p>\n<\/li>\n<\/ul>\n vi. Sustainability and long term operational implications<\/strong><\/p>\n Discuss how machine learning enabled risk prediction can support broader sustainability objectives, including accident reduction, optimised routing and more efficient allocation of inspection and enforcement resources<\/p>\n<\/li>\n<\/ul>\n vii. Methodological reflection and limitations<\/strong><\/p>\n Reflect on limitations and uncertainties related to data quality, model bias, generalisability and the difficulty of representing human and organisational factors in quantitative models<\/p>\n<\/li>\n<\/ul>\n Title page including module name, student ID and word count<\/p>\n<\/li>\n Abstract of 150\u2013200 words<\/p>\n<\/li>\n Introduction and problem definition<\/p>\n<\/li>\n System description and safety context<\/p>\n<\/li>\n Hazard identification and conceptual risk model<\/p>\n<\/li>\n Machine learning approach to risk prediction<\/p>\n<\/li>\n Human and organisational factors in ML supported operations<\/p>\n<\/li>\n Risk control options and cost benefit style assessment<\/p>\n<\/li>\n Sustainability and long term implications<\/p>\n<\/li>\n Methodological reflection and limitations<\/p>\n<\/li>\n References in Harvard style<\/p>\n<\/li>\n Appendices as required<\/p>\n<\/li>\n<\/ul>\n Word count of 3,000\u20134,000 words, excluding references, tables, figures and appendices, stated on the title page<\/p>\n<\/li>\n 11 or 12 point font, 1.5 line spacing and standard margins<\/p>\n<\/li>\n Harvard style referencing throughout<\/p>\n<\/li>\n Minimum of 12 high quality academic and industry sources<\/p>\n<\/li>\n Individual work only, with any use of AI tools acknowledged in line with institutional policy<\/p>\n<\/li>\n Submission via the designated VLE before the deadline<\/p>\n<\/li>\n<\/ul>\n On completion of this assessment, students will be able to:<\/p>\n Explain and critically evaluate machine learning methods for maritime risk prediction<\/p>\n<\/li>\n Integrate human and organisational factors into a structured maritime risk analysis<\/p>\n<\/li>\n Apply FSA-style reasoning to assess risk control options and their sustainability implications<\/p>\n<\/li>\n Communicate complex technical and human factors issues clearly to professional audiences<\/p>\n<\/li>\n Critically reflect on methodological assumptions, limitations and ethical dimensions of data driven safety management<\/p>\n<\/li>\n<\/ul>\n The assignment is marked out of 100 and contributes 40 percent of the module grade.<\/p>\n Problem definition and system description (15 percent)<\/strong> Hazard identification and risk model (20 percent)<\/strong> Machine learning approach and technical understanding (25 percent)<\/strong> Human and organisational factors integration (20 percent)<\/strong> Risk control options and sustainability perspective (10 percent)<\/strong> Use of literature (5 percent)<\/strong> Structure and academic writing (5 percent)<\/strong> Coastal administrations managing dense traffic corridors increasingly treat accident risk as a dynamic variable that can be estimated and updated in near real time rather than as a static reflection of historical statistics. When casualty records are combined with AIS tracks, bathymetry and meteorological data, supervised learning techniques such as random forests and gradient boosting can generate calibrated probability estimates for specific vessel types and locations. These estimates become operationally useful only when they align with how pilots, bridge teams and VTS operators perceive and act on risk cues, since algorithmic outputs that conflict with established mental models or appear at inappropriate moments are likely to be ignored or misinterpreted.<\/p>\n Formal Safety Assessment thinking emphasises that predictive accuracy alone is insufficient if risk information is not trusted, understood or acted upon by operators. In machine learning supported maritime safety systems, transparency of model logic and clear communication of uncertainty are critical for maintaining appropriate human engagement. Studies of data driven risk prediction show that interpretable models and carefully designed interfaces support better integration into operational decision making, reducing the likelihood of automation bias and strengthening the overall safety contribution of analytics based tools (Knapp and van der Hoorn, 2023).<\/p>\n Rawson, A. and Brito, M.P. (2022) Spatial modelling of maritime risk using machine learning. Risk Analysis<\/em>, 42(10), 2119\u20132137. Available at: https:\/\/eprints.bournemouth.ac.uk\/37810\/1\/Risk%20Analysis%20-%202021%20-%20Rawson%20-%20Spatial%20Modeling%20of%20Maritime%20Risk%20Using%20Machine%20Learning.pdf<\/a><\/p>\n Knapp, S. and van der Hoorn, I. (2023) Exploration of machine learning methods for maritime risk predictions. Erasmus University working paper. Available at: https:\/\/pure.eur.nl\/files\/156362752\/Exploration_of_machine_learning_methods_for_maritime_risk_predictions.pdf<\/a><\/p>\n Ziaul, M. et al. (2024) Predicting maritime accident risk using automated machine learning. Working paper. Available at: https:\/\/www.ziaulmunim.com\/wp-content\/uploads\/2024\/05\/Predicting-maritime-accident-risk-using-Automated-Machine-Learning-FINAL.pdf<\/a><\/p>\n Lin, Y., Zhang, F. and Yu, H. (2025) Machine learning applications for risk assessment in maritime transportation systems. Ocean Engineering<\/em>. Available at: https:\/\/khu.elsevierpure.com\/en\/publications\/machine-learning-applications-for-risk-assessment-in-maritime-tra\/<\/a><\/p>\nAssessment Context<\/h2>\n
Assessment Task<\/h2>\n
Task Description<\/h3>\n
\n
Core Requirements<\/h3>\n
\n
\n
\n
\n
\n
\n
\n
Indicative Structure<\/h2>\n
\n
Formatting and Submission Requirements<\/h2>\n
\n
Learning Outcomes Assessed<\/h2>\n
\n
Marking Criteria and Scoring Rubric<\/h2>\n
Clarity, realism and relevance of the defined system and safety problem.<\/p>\n
Depth and structure of hazard identification and quality of the conceptual risk model.<\/p>\n
Appropriateness and accuracy of the proposed ML approach, including data handling and evaluation.<\/p>\n
Quality of analysis showing how human and organisational factors influence both risk and ML tool use.<\/p>\n
Relevance and justification of proposed risk controls, including sustainability considerations.<\/p>\n
Range and critical engagement with academic and industry sources.<\/p>\n
Organisation, clarity and professional standard of writing and referencing.<\/p>\nLearning Resources<\/h2>\n