a specific marketing problem: the relationship between free shipping promotions and Internet shopping behavior. It develops original analytical models that generate hypotheses that are tested on Internet data. A data mining method called contrast set is used for empirical testing.
Existing Literature
The first essay relates to literature in the fields of market segmentation, pattern-based clustering, segmentation-based modeling, profiling, and signature discovery.
There are hundreds of clustering algorithms and segmentation approaches proposed in the statistics, data mining and marketing literature. There are distance based nonhierarchical clustering algorithms (e.g. k-means), hierarchical clustering algorithms (including agglomerative and divisive algorithms), model-based clustering algorithms and various other approaches that are not grouped into any of the above three categories (e.g. rule-based approaches, neural networks). The research that is most related to our method can be categorized into four groups that are now described.
Market Segmentation
Two dimensions of segmentation research include segmentation bases and methods. A segmentation basis is defined as a set of variables or characteristics used to assign potential customers to homogenous groups. Research in segmentation bases focuses on identifying effective variables for segmentation. Cluster analysis has historically been the most well-known method for market segmentation. Recently, much of the market segmentation literature has focused on the technology of identifying segments from marketing data through the development and application of finite mixture models. Finite mixture models allow customer behavior to be described by an appropriate statistical model with a mixture component. The main advantage of these models is that they