Employee Attrition Estimation Using Random Forest Algorithm

Today, almost all companies are concerned about retaining their employees. However, they are not able to recognise the real factors that make them quit their jobs. Many factors could be responsible for that (for example: cultural, financial, etc.). Each company has its way to treat its employees and assure their happiness. But often no measures are taken of the satisfaction rate. As a result, in many cases, employees quit their employment suddenly without an apparent reason. In the last decades, Machine learning (ML) techniques have gained popularity among researchers. It can propose solutions to a wide range of problems. Then, ML learning has the potential to make predictions to anticipate employee attrition. In this paper, the authors compare state-of-the-art solutions for the proposed machine learning algorithms using a real data set sample size of 1469. The results could be used to warn managers in order to change their strategies or behaviour. It could also be used to make recommendations to the managers to add some policies in order to retain their employees in the company. This study aims to present a comparison of different machine learning methods to give a prediction of employees who are likely to leave their company. The data set includes information about the current employees and the employees who had already quit their job with almost 50 valuable information units. This last combines many factors: social, cultural, financial, professional, and relational factors. Six different ML algorithms were used in this paper. Experimental results show that the Random Forest algorithm demonstrated the best capabilities to predict the employees' attrition. The best prediction accuracy was 85.12, that is considered as good accuracy.