Privacy protection and quantile estimation from noise multiplied data

Autor: Laura Zayatz, Tapan K. Nayak, Bimal K. Sinha
Rok vydání: 2011
Předmět:
Zdroj: Sankhya B. 73:297-315
ISSN: 0976-8394
0976-8386
DOI: 10.1007/s13571-011-0030-z
Popis: In this paper we address two inferential aspects of noise multiplied magnitude microdata. First, in the context of disclosure risk assessment of tabular magnitude data, we study the consequences of noise multiplication when an intruder tries to speculate a target unit’s value in a cell based on knowledge of the perturbed cell total and values of some units within the cell. This is related to some results in Nayak et al. (J Off Stat, 2011). Second, we develop Bayesian methods to infer about a quantile of a microdata set based on their noise perturbed values. Natural applications include estimation of quartiles and median of an original microdata set when only their noise perturbed versions are available.
Databáze: OpenAIRE