Decision Support in a Retail Chain: A Hybrid Approach Between Unsupervised Machine Learning and Multicriteria Decision Support

This paper presents a hybrid modeling for retail chain point-of-sale management, integrating unsupervised machine learning techniques with Multicriteria Decision Support (AMD). The non-hierarchical method of clustering K-means and the AMD AHP-Gaussian method were integrated. The proposed modeling was applied in a retail chain of the pharmaceutical segment with more than 70 points of sale in the city of Rio de Janeiro. Academically, the work contributes to the literature on hybrid modeling, while in the practical realm, it offers a strategic tool for retail chains, improving their operational efficiency and market competitiveness. The proposed modeling combines the power of machine learning, especially in identifying behavioral similarities between observations and variables in a database, with the effectiveness of AMD in evaluating and prioritizing them. This approach not only facilitates the decision-making process but also contributes to the development of more targeted and effective marketing strategies.