Evaluating Targeted Productivity in Bangladesh’s Garment Sector Using Machine Learning and Deep Learning with Explainable AI: A Data-Driven Method for Enhanced Production Planning

The garment industry significantly boosts Bangladesh’s economy, contributing over 80% to export earnings and employing millions of people. Effective pro- duction planning is vital for operational efficiency; however, establishing productivity targets often depends on manual judgment, resulting in unrealistic expectations. This study introduces a data-driven framework utilizing machine learning (ML) and deep learning (DL) methods to accurately predict productivity targets, supplemented by Explainable AI (XAI) tools such as SHAP and LIME. Various models—including Lin- ear Regression, Random Forest, KNN, XGBoost, Neural Network (NN), Deep Neural Network (DNN), and Convolutional Neural Network (CNN)—were assessed through 10-fold cross-validation. Performance metrics (MAE, MSE, RMSE, R2) identified the Neural Network as the top performer, achieving an R2of 0.987. The incorporation of XAI facilitated understanding of feature influences, emphasizing factors like SMV, WIP, idle time, and overtime. This methodology promotes transparent and intelligent planning, enabling factory managers to make better-informed, data-driven decisions.