Abstract:

This paper presents a quick data fusion algorithm (known as predictive isotonic fusion) that is customized on a case-by-case basis. The accuracy of this data fusion for target group ratings was compared against a commercially available syndicated data fusion. The authorrs found that there was no negative trade-offs from the much faster execution times; in fact, there were significant improvements in some cases. Furthermore, this data fusion method can accommodate many more predictor/matching variables which makes even larger improvements possible.