Molefe J Maleka

Abstract
The paper aimed to identify essential features that predict employee attrition at the multination consulting corporation. Losing talented employees is costly to organisations, as they incur recruitment and development costs and lose their competitive edge. Mobley, Griffeth, Hand and Meglino’s (1979) model was used to give theoretical grounding to the study. The study was quantitative, and the positivism paradigm influenced it. It used secondary data and bivariate analysis, including correlation and bagging (i.e., bagging classifier, random forest, decision tree, weighted bagging classifier, and decision tree estimator), to analyse data and decide which model best predicts attrition.
The data showed the top significant predictors of employee attrition: monthly income, overtime, daily rate, age, hourly rate, distance from home, number of companies worked and work-life balance. This study contributes by using the emerging ensemble method in predicting the features that predict employee attrition. Since 4 of the top 8 predictors are remuneration-related, managers should focus on revising and benchmarking best practices in the remuneration policy. Management should also focus on developing strategies to retain employees between 18 and 28 years.