Multi-criteria inventory ABC classification using Gaussian Mixture Model

Abstract ABC classification is a technique widely used by companies to deal with inventories consisting of very large numbers of distinct stock keeping units. Single-criterion ABC classification methods are often used in practice and recently multi-criteria methods have attracted the attention of academics and practitioners. Several models have been developed to deal with the multi-criteria ABC inventory classification (MCIC). To the best of our knowledge, very few researches have used the unsupervised machine learning methods to address the MCIC problem despite their attractive theoretical and practical properties. Therefore, in this paper, the Gaussian mixture model (GMM) is proposed to deal with the multi-criteria ABC inventory classification problem. GMM is a simple optimization model that can be used for classification purposes with a low computational time. A numerical investigation of the cost-service inventory of the GMM model is presented in this paper. The performance of the model is also compared to some mathematical programming-based MCIC models. The numerical study is conducted by means of a theoretical dataset, consisting of 47 stock keeping units, which has been commonly used in the literature. The numerical results show that the proposed model can have a promised performance along with the existing MCIC classification models in the literature in terms of the cost-service inventory efficiency.