Comparison of target selection methods in direct marketing

Target selection in direct marketing is an important data mining problem for which several modeling techniques can be used. Several data mining techniques have been applied in the last years to target selection. Logistic regression, neural networks, decision trees and fuzzy modeling methods are the most utilized techniques. However, they have never been explicitly compared up to know. This paper compares these techniques by using cross-validation measures. The four techniques are applied based on recency, frequency and monetary (RFM) value measures. The advantage of the RFM variables is that the customers’ behavior can be captured by using a relatively small number of features, which improves the transparency of the target selection models that are developed. The comparison using cross-validation measures is based on the application of target selection to mailing campaigns of a charity organization.