An analysis of brand interdependencies using Artificial Neural Networks

The purpose of this article is to present the abilities of Artificial Neural Networks in analyzing the existing structure of brand interdependencies compared to DE-MCI model. To achieve this pur-pose a comparative study is done based on POS data used by Cooper and Nakanishi in their monograph. The results suggest that ANN model outperform DE-MCI model with regards to model fit and they offer face valid estimates of self and cross-elasticities. Based on the transformed cross-elasticity estimates, a MDS map is produced. This competitive map is used to identify the existing interdepend-encies among the brands in the market.