Security vulnerabilities constitute a big problem for organizations, facing the need for a categorization system to handle these vulnerabilities. Moreover, because the number of new vulnerabilities is constantly increasing, manual categorization of them is a time-consuming task that is prone to errors. An alternative way to analyze vulnerabilities is to automate the process using machine learning (ML) algorithms. The purpose of the current research is to evaluate the performance of different ML algorithms for vulnerability categorization and prioritization based on the data obtained from a Danish company operating in the public transport sector. Overall, the experimental results indicate that the ML models performed effectively, demonstrating that ML is a suitable approach for automated vulnerability analysis. In this work, we present and evaluate two approaches: a hybrid model combining Random Forest (RF) with a Multi-Layer Perceptron (MLP), and a Learning to Rank (LtR) algorithm, LambdaRank. We also examine how class imbalance in real-world vulnerability data can affect model performance, and discuss broader implications of the results for designing more secure, resilient and trustworthy systems in related domains.