Selecting Relevant Clustering Variables in Mass Customization Scenarios Characterized by Workers’ Learning

Abstract

Clustering is an important technique in highly customized production environments, where a large variety of product models is typical. It allows product models with similar processing needs to be aggregated into families, increasing the efficiency of production programming and resources allocation. The quality of the clustering results, however, relies on using a set of relevant clustering variables. Our method selects the best clustering variables aimed at grouping customized product models in families. There are two groups of clustering variables: those generated by expert assessment on the features of products and those predicting the workers’ learning rate, obtained by means of learning curve modeling. The method integrates an elimination procedure with a k-means clustering technique. The method is illustrated on a shoe manufacturing process.