Coding: Statistical Data Masking Techniques
Aleksandra Slavković · Encyclopedia of Quantitative Risk Analysis and Assessment · 2008
Abstract Coding broadly refers to altering the data in some way. In computer science literature, it may refer to encryption, compression, or error correction. These are relevant for cryptography and cryptanalysis, and for data privacy and homeland security. In statistical confidentiality literature, coding encompasses various data‐masking techniques. These techniques generally introduce bias and variance to data. The question is how much and what kind of bias and variance is introduced and how this influences the utility of data‐mining algorithms and precision of identification, that is the false positive error rate. This article gives a brief overview of statistical disclosure limitation techniques for data masking, pointing out the likelihood of their use and potential impacts for data privacy.