| The development of privacy-preserving statistical methods has become increasingly necessary to protect sensitive personal information. In this thesis we combine Fisher's classification methods [1] with a perturbation method to protect privacy presented by Du et al. [2]. The result is a methodology for classification of an observation into one of two populations. The methodology is incorporated into R code and applied for illustrative purposes to a medical dataset. We also evaluate the misclassification rate associated with this technique. |
