Direct Marketing Response Models Using Genetic Algorithms

Direct marketing response models seek to identify individuals most likely to respond to marketing solicitations. Such models are commonly evaluatedon classification accuracyand some measureof fit-to-data. Givenlargecustomerfiles andbudgetarylimitations,only a fraction of the total file is typically selectedfor mailing promotional material.This desired mailing-depthpresents potentially useful information that is not consideredby conventional methods. This paper presentsa genetic algorithm basedapproachfor developingresponsemodels aimed at maximizing performanceat the desiredmailing depth.Here,depthof file informationis explicitly takeninto account during model development. Two modeling objectives,responsemaximizationat selectedmailing depth and fit-to-data,are consideredand tradeoffsamongstthese empiricallyexplored. Resamplingapproaches,effectivefor controllingoverfit to trainingdata,arealsoinvestigated.

• Type: Conference
• Year: 1998
• Where: KDD
• Authors: Siddhartha Bhattacharyya
• Keywords: Data Mining, KDD 1998, Marketing Response Models, Material.This Desired Mailing-depthpresents, Such Models