Statistical Probleme in Statistical Disclosure Control

点击翻译按钮获取中文摘要

Masaaki Sibuya and Hajime Yamato (2000, 2001a, 2001b), in collaboration with T. Nomachi, studied the properties of Pitman's model for the random partition of a finite set or that of a number. The maximum likelihood estimation of the parameters was found rather poor, and alternative metods were investigated. 'Its application to the statistical disclosure control is promising.Shido Sai (2000, 2001, 2002), in collaboration with A. Takemura, studied mainly nonparametric approach to the estimation of the number of individuals who are unique in both sample and population. If a decreasing size indexes (frequency spectrum or partition vector) in population is assumed, the maximum likelihood estimate is favorable although the computational work is combinatorial.A new technique for protecting private data is the Post Randomization Method, named PRAM, not the randomization during the survey. Tomoyuki Tarumi, with T. Fujino, is investigating the effect of the procedure to the multivariate statistical analyses. Manabu Iwasaki published a book on the introduction to the Incomplete Data Analysis. His analysis, with Y. Inaba, of the statistical table data on the aerial activities of commerce shows the published tables are vulnerable.