A Machine Learning Approach to Employee Performance Prediction within Administrative Information Systems

Abstract

This research project delves into the realm of employee performance prediction within organizational contexts, employing the k-Nearest Neighbors algorithm as the core predictive model. The central aim is to optimize organizational success by anticipating variations in individual performance metrics. Key features considered for prediction encompass the assessment of employee performance, average monthly work hours, and tenure with the company. Comparative analyses with alternative methodologies, such as k-Nearest Neighbors (KNN) algorithm, decision trees, and logistic regression, are conducted for a comprehensive evaluation. The dataset undergoes a meticulous 70-30 split for training and testing, resulting in a commendable accuracy of 97.32%. The implications of this research are significant for organizational management, providing a proactive tool to enhance and sustain employee performance, thereby fostering overall organizational success.