An Investigation of Employee Attrition Prediction Using Machine Learning Algorithms

Abstract

In an organization, employee attrition is a problem which is considered to be one of the most difficult to tackle. It is an established fact that, the employees are the major building blocks of any organization. Efficient employees add to the revenue, productivity, and efficiency of the company. But the companies face a major setback, when the employees start to leave the company, it rattles the HR people. It requires a bunch of dedicated and headstrong employees for any organization to flourish and produce effective results. So, it goes without saying that the employees are the most important factors in the growth of any company. The core of this project stems from the fact that, when employees leave, there is a significant reduction in the revenue, productivity, and efficiency of the company. To study upon this topic, we have extracted a dataset from Kaggle website comprising of 14,999 records with various parameters which influence the decision of employee attrition. Different types of machine learning algorithm have been used to predict the likeliness of employee attrition along with 10-folds cross validation to make sure the model is robust, and unbiased. 85 percent of the data has been used for training, and the rest is used for testing purpose. We find Random Forest has outperformed all other models with an accuracy of 98 percent, whereas Support Vector Machine has least accuracy of 60 percent.