Are the death rates in Game of Thrones realistic? A statistical case-study
Details
Abstract: Statistics is the science of learning from empirical data, and of measuring, controlling and communicating the associated uncertainty. Motivated by a common assertion about the pop culture phenomenon Game of Thrones (GoT), this talk will illustrate several important aspects of statistical analysis and data science. First, we will emphasise the importance of translating a vague question into a clear and testable hypothesis. Then, we will discuss the need to compare relevant and representative datasets, and how to find and collect such data. Finally, correct modelling of such data will be discussed, where we highlight the importance of dealing with censoring, and presenting some common methods in survival analysis. The chosen illustration for these insights is a dataset collected from the widely popular television show, Game of Thrones. The show is famous for the high mortality suffered by its main characters and its general violence. Some people even claim that the show is "too violent", and one may thus be tempted to investigate whether the wars in GoT really are bloodier than their historical inspirations. It is generally recognised that many elements in GoT bear some striking resemblances to the Wars of the Roses, a series of civil wars raging in England between 1455 and 1487, and we will thus compare the mortality rates in these wars with the mortality suffered by GoT characters.
Speaker: Céline Marie Løken Cunen (PhD candidate in statistics at the University of Oslo)