Maritime data surveillance has significantly grown in the last decade, notably through technologies like AIS (Automatic Identification System). Predicting vessel positions is crucial for various maritime applications, with a focus on trajectory forecasting. This involves anticipating vessel direction and future locations, which is vital for tasks such as search and rescue, traffic management, and pollution monitoring. Despite the abundance of AIS data available, predicting vessel paths remains challenging. AIS technologies, which use ship transponders, assist vessel traffic services (VTS) and serve as the foundation for tasks such as trajectory prediction and the classification of the arrival port given a route, namely ‘Port Classification’. AIS data includes essential vessel information, such as latitude, longitude, speed, and identity, transmitted via VHF signals. Deep learning models are considered state-of-the-art for analyzing AIS data. This study focuses on implementing a ‘Port Classifier’, evaluating several models including Conv1D, MLP and LSTM on AIS trajectories labelled through heuristic algorithms, and promising results are achieved, with Conv1D showing superiority in port classification tasks. Additionally, we conducted an Exploratory Data Analysis (EDA) to better understand the data. Our findings contribute to enhance maritime data analysis, and demonstrate potential applications for Official Statistics.