A Novel Protocol For Privacy Preserving Decision Tree Over Horizontally Partitioned Data

In recent times, there have been growing interests on how to preserve the privacy in data mining when sources of data are distributed across multi-parties. In this paper, we focus on the privacy preserving decision tree classification in multi-party environment when data are horizontally partitioned. We develop new and simple algorithm to classify the horizontally partitioned multi-party data. The main advantage of our work over the existing one is that each party cannot gather the other’s private data and it is simple and its performance is unmatched by any previous algorithm. With our algorithms, the computation cost and communication cost during tree building stage is reduced compared to existing algorithms.