Comparative Analysis of Machine Learning Models on Employees’ Attrition Prediction

Abstract

Employee attrition is the gradual decrease in the number of employees of an organization. This is a serious problem which can bring an entire organization down, mostly when the organization loses its most dedicated capable hands. One of the measures that an organization can take is to have a predictive system that can predict employees that may leave in order to find out the reason(s) and provide appropriate solution. This work provides a machine learning based interactive system that can reliably predict employee’s attrition. Data on attrition was obtained from Kaggle.com. Four Machine Learning Techniques – Decision Tree, Logistic Regression, Random Forest and Naïve Bayes were each used to generate a predictive model for attrition prediction. The performances of the models were tested using standard metrics and the most performing model was developed into an interactive web application. The models were developed using Python programming language while the web application was developed using Python, Hypertext Mark-up Language (HTML), Bootstrap and Django webserver. It is hopeful this system will be highly beneficial to organizations in order to make adequate prediction and planning in handling employee attrition.