Track: Inventory Control

Abstract

Inventory control is essentially a predominant decision to be undertaken by operations managers of a firm. A proper inventory control system ensures a sufficient amount of goods or materials to meet the firm's demand without facing undersupply or oversupply of materials. Traditionally, decision-makers classify inventory items into various classes or subgroups for easy monitoring and managing stock levels. The classification efficiency can be ameliorated by applying machine learning algorithms. The present paper focuses on developing a hybrid methodology that integrates machine learning algorithms with multi-criteria decision-making (MCDM) to facilitate multi-attribute inventory analysis. This technique enables the operator to leverage the benefits of both ML and MCDM. The data set is initially classified using MCDMs like the Simple Additive weighted (SAW) method and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) model. Performance metrics like overall cost and customer fill rate are utilized to rank the efficiency of the generated MCDM models. Supervised machine learning algorithms like Decision tree, Random forest, Support Vector Machine (SVM), XG boost and KNN are employed on the MCDM models, and the performance of the system is established in terms of the accuracy of the machine learning algorithm. In order to prevent the inventory model from inclining towards the majority class, up sampling of the dataset is undertaken. Analysing the results, it is concluded that the application of TOPSIS MCDM provides better results for the considered systems in both machine learning and non-machine learning performance indices.

Keywords

ABC classification, Multi criteria decision making, Machine learning, Resampling.