Apply XGBoost, Random Forest, and Neural Netwrk (MLP) Models for Predicting Employee Absenteeism Using Workforce Data Analytics

Abstract

Absence issues among employees create numerous problems for workplace productivity and operational continuity. The paper analyzes machine learning approaches to predicting employee absenteeism based on controlled workforce information. Scientists applied a thorough preprocessing pipeline to a realistic dataset, which involved dealing with missing values as well as encoding categories along with feature numerical scaling. An exploratory analysis of the data and feature importance evaluation helped identify proper predictive attributes.
Random Forest, together with XGBoost and Multi-Layer Perceptron (MLP) Neural Network, served as the classification models to create predictive models, which were assessed using accuracy and precision and recall, and F1-score metrics. XGBoost proved its superiority as a predictive model by reaching 81.03% accuracy during testing while showing balanced results. This study highlights that ensemble-based machine learning approaches present opportunities to boost data-driven decision-making for human resource management operations. The coming paper will concentrate on adding more content to datasets as well as improving model visibility while studying adaptive learning methods to enhance real-world systems and stability.