Using machine learning to predict employee turnover

: a case study of the Willis Towers Watson Lisbon Hub

Abstract

Nowadays, the business world is characterized by high competitiveness, where every factor is essential for ensuring the success of a company. In this sense, Human Resources take on great importance, being the essential component for ensuring the success of any organization. Thus, retaining highly qualified employees is one of the biggest challenges faced today, with employee turnover being a costly and destructive problem for companies. Thee Lisbon Regional Delivery Hub (LRDH) of Willis Towers Watson (WTW) is no exception, with its highly qualified employees leaving annually. This thesis has a dual purpose: to investigate the factors that influence employee turnover at Willis Towers Watson and to determine the most effective predictive model for predicting that turnover. This study found that employees with longer tenure, those who have been promoted, Portuguese nationals, and individuals in higher-level positions are more likely to remain active in the organization. Additionally, the study also showed the effectiveness of machine learning models, with the random forest model being highly capable of predicting employee turnover at WTW. These results provide valuable insights for the company’s human resources department, as it is now possible to identify the factors that most influence employee departures and, thus, adopt retention measures. By addressing these factors, the LRDH of WTW can not only reduce employee turnover, but also improve its productive and financial performance.

Date of Award	25 Jan 2024
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Pedro Afonso Fernandes (Supervisor)