Método Cluster-RNA para Clasificar, Caracterizar y Pronosticar Perfiles Competitivos del Sector Tiendas Minoristas en la Ciudad de Barranquilla

Abstract

Objective– To develop a method to classify, characterize and forecast competitive profiles of the retail stores sector based on the integration of the cluster analysis technique and artificial neural networks.

Methodology– For the above, the literature related to the competitiveness of retail stores was reviewed, from which variables associated with this research were identified. The information analyzed corresponds to 224 retail stores in the city of Barranquilla.

Results– The cluster analysis allowed to characterize 4 competitive profiles of the sector that showed to be homogeneous intragroup and heterogeneous extragroup. The artificial neural network model showed a 91.3% correct classification in the reserve sample, which inferred the capacity of classification of the network model and the discriminant capacity of the variables related to the knowledge of products and prices, the practices of inventory and sales, presence in the market, differentiated attention, location and variety of products in the identified profiles.

Conclusions– The results of the research show high capacity of the cluster-RNA method, to classify and project competitive profiles from which improvement processes can be designed.