Exploiting Data Spatial Dependencies for Employee Turnover Prediction

Sandra Maria Pereira, Jéssica da Assunção Almeida de Lima, Alessandro Vieira, Wladmir Brandão, Wladmir Brandão

2024

Abstract

Machine learning techniques have been increasingly employed to address problems within the field of human resources. A significant issue in this domain is predicting employee turnover, related to the probability of an employee leaving the company. Employee turnover is directly related to the availability of knowledge and resources that affect the continuity of the company’s goods and services supply. Managing employee turnover involves multiple areas of expertise, rendering it a complex problem. This article proposes a methodology to determine whether prediction problems exhibits spatial dependence, thereby demanding the use of spatial models over non-spatial models for optimal resolution. Experimental results show that significant differences arise when analyzing correlations that consider the geographical positioning of the data. Particularly, prediction models that use geographic features to predict employee turnover outperform prediction models that do not use them, with gains ranging from 9.6% to 19.6% in the standard deviation of MAPE, from 5.5% to 10.4% in MAE, and from 0.99% to 2.9% in RMSE.

Paper Citation

in Harvard Style

Maria Pereira S., da Assunção Almeida de Lima J., Vieira A. and Brandão W. (2024). Exploiting Data Spatial Dependencies for Employee Turnover Prediction. In Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST; ISBN 978-989-758-718-4, SciTePress, pages 250-257. DOI: 10.5220/0012948600003825

in Bibtex Style

@conference{webist24,
author={Sandra Maria Pereira and Jéssica da Assunção Almeida de Lima and Alessandro Vieira and Wladmir Brandão},
title={Exploiting Data Spatial Dependencies for Employee Turnover Prediction},
booktitle={Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST},
year={2024},
pages={250-257},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012948600003825},
isbn={978-989-758-718-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Web Information Systems and Technologies - Volume 1: WEBIST
TI - Exploiting Data Spatial Dependencies for Employee Turnover Prediction
SN - 978-989-758-718-4
AU - Maria Pereira S.
AU - da Assunção Almeida de Lima J.
AU - Vieira A.
AU - Brandão W.
PY - 2024
SP - 250
EP - 257
DO - 10.5220/0012948600003825
PB - SciTePress