Publicação
Identifying and characterizing employee groups by turnover risk using predictive analytics
| Resumo: | This project presents a predictive analytics project developed in a European multinational to understand and predict the turnover of its employees. It analyses the Human Resources current challenges, such as the increasing global competition for talent, where players compete for scarce skillsets such as technology and data science, and the new strategies necessary to deal with this scenario. The study explores the literature review of these contextual matters and of the studies of variables that influence turnover, generating insights and input for applying techniques aligned with the new mindset of identifying ‘flight-risk’ groups and developing targeted actions instead of only one-size-fits-all solutions. The project gathered data from different sources of the organization, designed variables, based on a literature review and internal brainstorms, treated data quality issues, transformed the data and applied three different machine learning algorithms to develop a classification predictive model. The study evaluated 46 input variables and selected a set of 26 that had higher impact on the turnover which were used in the models. Finally, it applied clustering techniques to divide employees in clusters, and identified two containing more extreme turnover behaviors (“Loyal” and “Flight risk”) and described them accordingly to their main characteristics contributing with practical insights to support potential decisions. |
|---|---|
| Autores principais: | Vidotto, Bruno Cassanta |
| Assunto: | Turnover Attrition Human Resources People Analytics Machine Learning Predictive Analytics |
| Ano: | 2021 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | This project presents a predictive analytics project developed in a European multinational to understand and predict the turnover of its employees. It analyses the Human Resources current challenges, such as the increasing global competition for talent, where players compete for scarce skillsets such as technology and data science, and the new strategies necessary to deal with this scenario. The study explores the literature review of these contextual matters and of the studies of variables that influence turnover, generating insights and input for applying techniques aligned with the new mindset of identifying ‘flight-risk’ groups and developing targeted actions instead of only one-size-fits-all solutions. The project gathered data from different sources of the organization, designed variables, based on a literature review and internal brainstorms, treated data quality issues, transformed the data and applied three different machine learning algorithms to develop a classification predictive model. The study evaluated 46 input variables and selected a set of 26 that had higher impact on the turnover which were used in the models. Finally, it applied clustering techniques to divide employees in clusters, and identified two containing more extreme turnover behaviors (“Loyal” and “Flight risk”) and described them accordingly to their main characteristics contributing with practical insights to support potential decisions. |
|---|