| Popis: |
Advances in information technology (e.g., scanner data, cookies, and other electronically based data collection methodologies) have enabled researchers to collect unprecedented amounts of individual-level customer data. As a result, customer databases are becoming increasingly larger and more complex, and may tax the capabilities and exacerbate the shortcomings of the techniques currently used to analyze them. To address this challenge, we examine the use of artificial neural networks (ANNs) as an alternative means of segmenting retail databases. In particular, we investigate the Hopfield–Kagmar (HK) clustering algorithm, an ANN technique based on Hopfield networks, and empirically compare it to K -means and mixture model clustering algorithms. Our results indicate that ANNs may be more useful to retailers for segmenting markets because they provide more homogeneous segmentation solutions than mixture model and K -means clustering algorithms, and are less sensitive to initial starting conditions. |
Full Text from ScienceDirect