Predicting kills in Game of Thrones using network analysis

TV series such as HBO’s Game of Thrones have a high number of dedicated followers, mostly due to the dramatic murders of the most important characters. In our work, we try to predict killer and victim pairs using data on previous kills as well as additionalmetadata. We construct a network where two character nodes are linked if one killed the other, then use a link prediction framework to evaluate different techniques for kill predictions. Lastly, we compute various network properties on a social network of characters anduse them as features in conjunction with classic data mining techniques. Due to the small size of the dataset and the somewhatrandom kill distribution, we cannot make accurate predictions with standard indices alone, although using them in conjunction with additional rules based on degrees has yielded results that are more reliable. The features we compute on the social network help theclassic machine learning approaches; however, they do not yield very accurate predictions. The best results overall are achieved using indices that use simple degree information, the best of which result in the Area Under the ROC Curve of 0.875.