HR Analytics: Analysis of Employee Attrition Using Perspectives from Machine Learning

Abstract

Over the past few decades, technology has transformed nearly every phase of our lives. One of the most important transitions has happened in the area of analytics lending unimaginable power in the hands of business people to transform themselves by leveraging information. With the growing interest in machine learning (ML) technology businesspeople are interested in exploring its use in business practices. This is because they have a singular mission of gaining a competitive advantage. Employee attrition is one of the largest and most unknown costs an organization may have to face. This chapter provides an extensive overview of employee turnover using ML techniques. The prediction is completed utilizing the information sourced by IBM Analytics. The real dataset consists of 35 attributes or features and 1470 samples. In this chapter, different classification techniques are used for predicting employee attrition. The results obtained after model selection are expressed in terms of the confusion matrix along with the algorithm that processes the optimum results. The random forest algorithm is found to have delivered the optimum results for the provided dataset. The outcome of the research found seven variables as critical driving factors that contributed to employee attrition. The results of the algorithm give 100% accuracy, precision, recall rate, and specificity while giving an 85.228% CV score and ROC score of one. This study will give senior-level executives and policymakers a clear viewpoint from which they can decide whether to keep a majority of the employees within the organization. Future studies could enhance the analysis by taking into account additional elements that have a beneficial impact on employee attrition rates, such as poor hiring practices, a hostile workplace environment, as well as a lack of feedback and appreciation.