IMPLEMENTATION OF MACHINE LEARNING ON EMPLOYEE ATTRITION BASED ON PERFORMANCE PARAMETERS USING PARTICLE SWARM OPTIMIZATION AND ENSEMBLE CLASSIFER METHODS

This research aims to apply machine learning to predict the start of employee attrition by considering performance parameters and other related factors in the company environment. Employee attrition refers to employee turnover in an organization for various reasons such as resignation, moving, retirement, and so on. This research uses a dataset originating from the IBM HR Analytics Employee Attrition dataset available on Kaggle (https://www.kaggle.com/) which consists of 35 attributes. Particle Swarm Optimization (PSO) method is a dimension reduction method to improve the efficiency and performance of machine learning models by reducing unnecessary data. The machine learning approaches used in the early prediction of employee attrition in this research are Support Vector Machine, Deep Learning and Neural Network methods. This research will combine the dimensionality reduction process with machine learning to obtain employee attrition prediction results that are optimized using the Ensemble method, namely Bagging and Boosting to increase the accuracy value of the prediction results. The results of this research show that applying dimensionality reduction using the PSO method can improve the accuracy of results on the IBM HR Analytics Employee Attrition dataset. The best accuracy in attrition prediction was obtained by the Deep Learning method with an accuracy value of 86.94%, a precision value of 88.90%, and a recall value of 96.40% after combining it with PSO and optimizing with Bagging.