Learning criteria weights with TOPSIS method and continuous VNS for multi-criteria inventory classification

Abstract

Listen

Translate article

The ABC analysis is a well-known methodology used in inventory management to classify inventory items into three predefined and ordered categories A, B and C: category A contains the most valuable items and category C contains the least valuable ones. The aim of this analysis is to keep - by focusing on the few critical items (of category A) - related inventory costs under control within a supply chain. According to ABC analysis, the classification of items into one of the above categories is based on their weighted scores. The score of each item is the result of an aggregation function that combines the item evaluations on the different criteria and the criteria weights. In this paper, we propose an Automatic Learning Method (ALM) that infers the criteria weights in order to produce a classification of items that minimizes an inventory cost function. The proposed ALM is based on TOPSIS method (Technique for Order Preference by Similarity to Ideal Solution) to compute the score of each item and a Continuous Variable Neighborhood Search (CVNS) to infer the criteria weights. To test the performance of the proposed ALM with respect to some others ABC inventory classification models, a benchmark data set of 47 items from an Hospital Respiratory Therapy Unit is used.