Music streaming companies seek to increase enhance the user experience by offering personalized music recommendations. Moreover, users value personalization as a top feature on a streaming service. Music preferences can be represented as a dynamic graph of users interacting with music genres over time. Our goal is to predict the music preference of a user using classical graph algorithms, statistical inference and Temporal Graph Neural Networks. We will work with the Temporal Graph Benchmark for our study and if possible we will apply our models to other real world networks.