A DATA-DRIVEN FRAMEWORK FOR LEAN PRODUCTION OPTIMIZATION IN SAP-INTEGRATED MANUFACTURING SYSTEMS
TL;DR
The proposed Gradient Boosting model substantially outperformed these baseline algorithms, achieving 99.99% accuracy, precision, recall, and F1-score and demonstrates the potential of ensemble learning to identify complex manufacturing patterns and support scalable, accurate, and real-time lean production optimization in SAP-enabled smart manufacturing environments.
Abstract
The increasing adoption of intelligent manufacturing and Industry 4.0 technologies has intensified the need for intelligent production planning and real-time operational optimization within enterprise systems. SAP Production Planning (SAP PP) and SAP Digital Manufacturing enable effective integration between production planning and shop-floor execution. However, optimizing lean production requires advanced data analytics and a reliable model for classifying operational efficiency. In this study, feature-engineering techniques were applied to an intelligent manufacturing dataset, producing 32 relevant attributes. The data were processed through outlier removal, categorical encoding, feature standardization using StandardScaler, and stratified cross-validation to ensure a robust and balanced evaluation. Comparative results showed that the Decision Tree, Naive Bayes, and Multilayer Perceptron models achieved accuracies of 94%, 91.14%, and 92%, respectively. The proposed Gradient Boosting model substantially outperformed these baseline algorithms, achieving 99.99% accuracy, precision, recall, and F1-score. These findings demonstrate the potential of ensemble learning to identify complex manufacturing patterns and support scalable, accurate, and real-time lean production optimization in SAP-enabled smart manufacturing environments.