In recent years, the application of deep learning techniques to enhance the accuracy of vessel trajectory prediction, using AIS satellite data, has garnered significant attention within the scientific community. Vessel trajectory prediction plays a pivotal role in maritime operations, facilitating safe and efficient navigation in bustling waterways. The behavior of vessels at sea is inherently diverse and contingent on their respective types. This data, which is predominantly quantitative in nature, is invaluable for probing intricate phenomena, behaviors, and tendencies. The primary challenge lies in effectively integrating this wealth of information into the calculations performed by deep learning models. In this chapter, we delve into the potential advantages of incorporating deep learning algorithms alongside categorical vessel-type data for the purpose of enhancing vessel trajectory prediction, thereby bolstering maritime safety. Our experimental approach entails the utilization of various categorical encoding techniques, including ordinal, one-hot, and multidimensional Embedding, to enable a comparative assessment of prediction accuracy in scenarios where categorical vessel-type features are not employed. Across all these techniques, we leverage Long Short-Term Memory models for recursive multistep forecasting, advancing our understanding of this critical domain.