Privacy preserving associative classification on vertically partitioned databases

The growing needs of multiple parties interaction in corporate and financial sector emphasize the need of developing privacy preserving and efficient distributed data mining algorithms. Even though a lot of research work is progressing in this area to transform efficient centralized mining models to work on horizontal and vertical partitioned databases there is lack of associative classification model that can perform classification on vertically partitioned databases. In order to overcome such needs this paper proposes an associative classification model on vertically partitioned databases. By considering privacy requirements in case of data sharing among multiple parties a scalar product based third party privacy preserving model adopted for proposed model. The proposed model accuracy tested on UCI data bases given encouraging results.