Allocation of catalogs to collective customers based on semiparametric response models

Most approaches to the problem of catalog allocation focus on the decision to send one catalog to an addressee. In practice there is the important group of “collective customers”, who submit catalogs to acquaintances to collect their orders. Therefore collective customers should often obtain more than one catalog. We propose a profit maximization approach based on semiparametric generalized additive models (GAM) for these customers' responses to the number of allocated catalogs. Response variables considered are purchase, monetary value of purchase, returning goods and monetary value of returned goods. An exchange algorithm is developed and implemented to maximize the firm's profit given a fixed total number of catalogs.