An Explicative and Predictive Study of Employee Attrition using Tree-based Models

10 Pages Posted: 13 Jun 2019

See all articles by Nesreen El-rayes

Nesreen El-rayes

New Jersey Institute of Technology

Michael Smith

New Jersey Institute of Technology - Martin Tuchman School of Management

Stephen Michael Taylor

New Jersey Institute of Technology

Date Written: June 1, 2019

Abstract

We develop tree-based models to estimate the probability of an employee leaving a firm during a job transition from a dataset of anonymously submitted resumes through Glassdoor’s online portal. Dataset construction and summary statistics are first summarized followed by a more in depth examination through four exploratory studies. Insights provided by these studies are then used to engineer features that serve as input into subsequent attrition related predictive models. We finally perform a thorough search through several dozen binary classification techniques in the cases of an original and extended feature set. We find tree-based methods including random forests and light gradient boosted trees provide the overall strongest predictive performance. Finally, we summarize ROC curves for several such models and describe future potential research directions.

Keywords: Attrition, Human Resources, Gradient Boosted Trees

JEL Classification: J630

Suggested Citation

El-rayes, Nesreen and Smith, Michael and Taylor, Stephen Michael, An Explicative and Predictive Study of Employee Attrition using Tree-based Models (June 1, 2019). Available at SSRN: https://ssrn.com/abstract=3397445 or http://dx.doi.org/10.2139/ssrn.3397445

Nesreen El-rayes

New Jersey Institute of Technology ( email )

University Heights
Newark, NJ 07102
United States

Michael Smith

New Jersey Institute of Technology - Martin Tuchman School of Management ( email )

United States

Stephen Michael Taylor (Contact Author)

New Jersey Institute of Technology ( email )

University Heights
Newark, NJ 07102
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
52
Abstract Views
261
rank
390,693
PlumX Metrics