TGEM (Temporal Graph Embedding) is a project created by a group of 3 Data Science students. The team is investigating the graph embedding methods - encoding their structure into a vector of features, which can then be analyzed using the majority of known machine learning algorithms (including neural networks). Particular emphasis is placed on temporal graphs, i.e. those whose structure changes over time (e.g. social networks). Embedding of graphs is useful due to the presence of large amounts of graphical structure data and difficulties in processing such data in an unchanged form. The issues studied have a wide range of practical applications, including reduction of computational costs, connection prediction (e.g. recommendation systems - suggesting friends on social networks) or classification of nodes (e.g. defining the characteristics of a person based on knowledge about their circle of friends).