CoCAT: a new similarity coefficient for solving the machine-part cell formation problem

This paper presents a new similarity measure (CoCAT) for finding the machine-part cells from the machine-part incidence matrix. When compared with the non-Jaccardian similarity measures, CoCAT provides a positive similarity value even if one common attribute is found between the two parts/machines, thus facilitating efficient cluster formation for the sparse matrices. This paper presents the performance of the CoCAT similarity measure with various clustering algorithms. The proposed similarity measure has enhanced the performance of the existing heuristic approaches and achieved a higher grouping efficacy (GE) for 6% of the standard test instances, and equal to the best-in-class GE for 94% of the standard test instances. A higher benchmark GE value for the dataset 33 and 34 has been achieved in the context of machine-part cell formation without singletons.