A few weeks ago I finally signed up for Kaggle and got my feet wet with a little machine learning. In this project, I analyzed (simulated) Human Resources data with respect to 14,999 employees to predict (and understand) which employees would give their two weeks notice. Employee retention (or conversely, 'churn') is a key problem faced by companies, as it is significantly more expensive to find, hire, and train new employees than it is to retain current ones. Thus many (all?) employers have a clear interest in understanding why people tend to leave and to identifying those who are currently at the highest risk of leaving.

Continue reading