Evolutionary neural classification approaches for strategic and operational decision support in retail store planning

In the domain of classification tasks, artificial neural nets (ANNs) are prominent data mining methods. Paradigms like learning vector quantization (LVQ) and probabilistic neural net (PNN) are suitable classifiers. In this paper, new approaches of evolutionary optimized LVQs and PNNs are proposed. Their classification accuracy is compared with results of standard PNN and LVQ. The complex real-world scenario includes planning of retail stores. Branch locations are classified in terms of revenue and profit. Results are based on data reflecting external infrastructure and internal aspects of existing branches. They support decisions about establishing, modifying or closing down a store.