Microaggregation Heuristic Applied to Statistical Disclosure Control

Abstract The dissemination of microdata by public institutions, especially National Statistical Offices (NSOs), and data sharing between public and private organizations and the academia are of undeniable importance today. However, very important ethical and legal aspects related to privacy protection, whether for an individual or a company, often restrict this cooperation. In this sense, the methods of Statistical Disclosure Control (SDC), or, more specifically, microaggregation techniques, offer an alternative to enable the dissemination and sharing of microdata through the control of disclosure risk. In this article, we propose a new heuristic method that adopts concepts from Biased Random Key Genetic Algorithm (BRKGA) metaheuristic and present a comparative study with well-known microaggregation methods. Computational simulations show that the proposed method consistently outperforms other methods in terms of simultaneously reducing information loss and disclosure risk. In addition, the proposed method allows the adopting or even combining of any information loss and disclosure risk metrics. This feature delivers great flexibility and much more control to the privacy protection process.