An Integrated Machine Learning Framework for Enhanced Performance Evaluation of Garments Manufacturing Operators

Abstract

Employee performance evaluation plays a crucial role in determining key career outcomes such as wage increments, bonuses, promotions, and terminations. In the garments manufacturing sector, where efficiency and productivity are vital, accurate and objective assessment of worker performance is essential for maintaining both quality and fairness. Aim of this study is to developing a data-driven approach to classify workers based on their performance using modern machine learning techniques. The study employs the K-Means clustering algorithm, an unsupervised learning technique, to group workers into meaningful categories according to multiple performance-related variables. The dataset, collected from Dewhirst Shanta Group, Dhaka EPZ, consists of 550 garment operators. The performance indicators considered include age, attendance percentage, efficiency rate, experience (in months), productivity (per day pieces), and daily wages. Data preprocessing was performed to ensure consistency and quality before analysis. Based on the K-Means clustering results, three distinct performance groups were identified: great (high-performing), good (moderate-performing), and average (medium-performing). The Silhouette Score obtained was 0.197, indicating fair clustering quality with moderate separation between groups. Among all operators, 167 workers were categorized as high-performing, 176 workers as moderate-performing, and 207 workers as average-performing.
The findings provide valuable insights for management and policy decision-making in the garments industry. Employers can utilize these results to optimize human resource allocation, design fair incentive systems, and plan targeted skill development programs. Similarly, employees can gain a clearer understanding of their performance standing and improvement areas. Overall, this study demonstrates that integrating machine learning-based clustering techniques can enhance the accuracy, fairness, and effectiveness of performance evaluation systems in labor-intensive industries like garments manufacturing.