Customer Response Model in Direct Marketing: Solving the Problem of Unbalanced Dataset with a Balanced Support Vector Machine

Customer response models have gained popularity due to their ability significantly improve the likelihood of targeting customers most likely buy a product or service. These are built using databases previous customers’ buying decisions. However, smaller number in these often bought service than those who did not do so, resulting unbalanced datasets. This problem is especially significant for online marketing campaigns when class imbalance emerges many website sessions. Unbalanced datasets pose specific challenge data-mining modelling inability algorithms capture characteristics classes that unrepresented dataset. paper proposes an approach based on combination random undersampling and Support Vector Machine (SVM) classification applied dataset create Balanced SVM (B-SVM) data pre-processor analysed with several classifiers. The experiments indicate B-SVM strategy combined methods increases base models’ predictive performance, indicating efficiently pre-processes data, correcting noise imbalance. Hence, companies may use more select respond campaign.