Utilizing 3D models for the Prediction of Work Man-Hour in Complex Industrial Products using Machine Learning

The integration of machine learning techniques in industrial production has the potential to revolutionize traditional manufacturing processes. In this study, we examine the efficacy of gradient-boosting machine learning models, specifically focusing on feature engineering techniques, applied to a novel dataset with 3D product models pertaining to work moan-hours in metal sheet stamping projects, framed as a regression task. The results indicate that LightGBM and XGBoost surpass other models, and their effectiveness is further enhanced by employing feature selection and synthetic data generation methods. The optimized LightGBM model exhibited superior performance, achieving a MAPE score of 10.78%, which highlights the effectiveness of gradient boosting mechanisms in handling heterogeneous data sets typical in custom manufacturing. Additionally, we introduce a methodology that enables domain experts to observe and critique the results through explainable AI visualizations.