A Comparative Statistical Analysis of Decision Tree and AdaBoost Ensemble for Employee Performance Classification in the Hospitality Sector

Abstract

This study aims to classify the factors affecting employee job satisfaction in Hotel X using the Decision Tree (DT) and Adaptive Boosting (AdaBoost) methods. The hospitality industry relies heavily on human capital to deliver high-quality services, and employee satisfaction is directly linked to service excellence, loyalty, and organizational performance. Data were collected from Hotel X’s internal Employee Satisfaction Index (ESI), comprising 70 records and 9 response indicators across multiple departments. Exploratory Data Analysis (EDA), correlation analysis, and label encoding were performed to prepare the dataset. The Decision Tree was first utilized to model the classification of employee satisfaction levels, followed by optimization using the AdaBoost ensemble method to enhance predictive accuracy. Three simulations were conducted using training-to-testing ratios of 70:30, 75:25, and 80:20, respectively. The results show that AdaBoost consistently improved the classification performance, achieving the highest accuracy of 93% in the third simulation. These findings underscore the significance of ensemble learning techniques in enhancing model reliability for human resource analytics in the hotel industry. This research demonstrates the practical value of combining DT and AdaBoost for workforce data analysis in service-based organizations. The model can be adapted to various industries that prioritize employee satisfaction as a key performance driver.