Secure” log-linear and logistic regression analysis of distributed databases

Abstract. The machine learning community has focused on confiden-tiality problems associated with statistical analyses that “integrate ” data stored in multiple, distributed databases where there are barriers to sim-ply integrating the databases. This paper discusses various techniques which can be used to perform statistical analysis for categorical data, especially in the form of log-linear analysis and logistic regression over partitioned databases, while limiting confidentiality concerns. We show how ideas from the current literature that focus on “secure ” summa-tions and secure regression analysis can be adapted or generalized to the categorical data setting.