Revolving drivers: data mining and discovering the causes of driver turnover

Abstract

Turnover (or Churn) is a great concern to every industry. Employees who leave represent hours of training wasted and the expense of hiring a replacement, something undesirable for any business. Few industries experience the problem as acutely as the trucking industry, where turnover rates have been as high as 90%. Uncovering the underlying reasons that are behind why so many drivers leave their jobs is a point of priority for many trucking companies. A solution, or even an explanation, could mean hundreds of training hours and thousands of dollars saved. In this M.Sc. thesis, I examine real-life data from a trucking company and use a random forest model to understand the driver turnover situation.