Processing Time Estimation in the Textile Warp Preparation Shop with Machine Learning Approaches

Accurate determination of processing time has critical importance in the production planning process for several reasons, such as effective planning of production resources and ensuring customer satisfaction by meeting due dates. Traditionally, processing time is often determined through time studies or simple calculations and is usually assumed to be known prior to planning. However, in some special cases, processing time varies depending on many parameters. Such a situation occurs in the warp preparation process, which is one of the important steps of woven fabric production. In this study, supervised machine learning approaches were used to estimate the warp preparation process time based on data obtained from the ERP system of the enterprise where the application was implemented. Twelve different supervised machine learning algorithms were applied to both training and test datasets, and the results are presented comparatively. It was observed that boosting algorithms outperform others in terms of both training/tuning time and estimation accuracy.