A holistic data analytics approach to ship inspection reporting

Abstract

Maritime inspection analysis has become an emerging topic in recent years, as practical solutions are sought to improve the pre- and post-inspection analysis in shipping operations. With a focus on finding practical solutions to enhance the pre- and post-inspection process in shipping operations, such as The Oil Companies International Marine Forum (OCIMF) Ship Inspection Report Programme (SIRE), RightShip, The Tanker Management Self-Assessment (TMSA), Chemical Distribution Institute (CDI), there is a growing demand for effective methodologies. The objective of this research is to enhance this field by examining documented observations through the utilization of both natural language processing (NLP) and machine learning (ML) methods. The main goal of this study is to make a valuable contribution to this field by analyzing reported observations. This will be accomplished by employing a combination of natural language processing (NLP) and machine learning (ML) techniques. Additionally, a statistical algorithm model will be utilized to conduct analysis using demographic data. To achieve the objectives of the study, a robust methodology was developed, which leverages the benefits of the American Bureau of Shipping Maritime Root Cause Analysis Tool (ABS-MARCAT). This tool enables the systematic initiation of a potential causes database, incorporating a substantial number of 2383 observations. By employing ABS-MARCAT, the study aims to provide a comprehensive foundation for analyzing and understanding the causes behind reported observations and determining corrective and preventive action tips for elimination of this causes. One of the key contributions of this research is the development of an NLP-based ML algorithm. This algorithm plays an important role in predicting the causes of new entries and determining corrective and preventive action tips in the inspection report's observations. The algorithm's performance demonstrates high accuracy, with results varying between 0.90 and 0.98 across different causation categories. Such accuracy is promising, as it allows for effective identification and classification of causes, providing valuable insights for decision-making in the maritime industry. Another important contribution of this research is the statistical algorithm model that can produce frequencies of causes based on independent variables such as ship name, inspector name, oil major company name and port name. The statistical algorithm model provides predictions about the areas to be considered according to the information required before the inspection. By presenting the frequencies of the cause categories according to the independent variables, it provides a decision support system in the process of predicting the inspection parts to be considered before the inspection. Another important contribution of this research is to suggest corrective and preventive action tips to eliminate the causes of the observations after the causes are identified. The corrective and preventive action tips determined by maritime experts will add a different dimension to the decision-making processes by providing solution suggestions after the analysis of the inspection reports. The pre- and post-inspection analysis model developed in this study holds great potential for enhancing fleet safety and efficiency. By providing maritime executives with an accurate tool to analyze inspection data, it enables them to make informed decisions and take proactive measures to address potential issues. The model serves as a third-party solution for the shipping industry, offering an independent and reliable means of analyzing and assessing inspection data. Looking ahead, future studies are planned to further refine and expand this model. The aim is to conceptualize it as a platform as a service (PaaS) offering, which would enable wider access and utilization by stakeholders in the maritime industry. By transforming the model into a PaaS, it has the potential to become a valuable resource for multiple organizations, facilitating improved fleet safety, operational efficiency, and informed decision-making. In conclusion, this study addresses the emerging field of maritime inspection analysis by developing a robust pre- and post-inspection analysis model. Through the integration of statistical algorithm model, NLP, ML, and the MARCAT tool, the study offers a holistic approach to analyzing reported observations and statistical data. With its high accuracy, the model has the potential to make a significant contribution to the improvement of fleet safety and efficiency. Furthermore, by conceptualizing it as a platform as a service, the study paves the way for wider adoption and application of the model within the shipping industry.