Whilst at Viewerslogic, I led a project looking at our panellists data and identifying key features that would predict whether they churned after a number of weeks. This heavily involved feature engineering and presenting decision trees to non-technical stakeholders. I have included relevant research for improving this project.