2005 Fiscal Year Final Research Report Summary
Studies on disclosure control of microdata and statistical analysis of disclosed data
| Project/Area Number |
14208023
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Statistical science
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
TAKEMURA Akimichi The University of Tokyo, Graduate School of Information Science and Technology, Professor, 大学院・情報理工学系研究科, 教授 (10171670)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SHIBUYA Masaaki Takachiho University, School of Commerce, Professor, 商学部, 教授 (20146723)
SAI Shidou Okayama Shoka University, Departments of Economics, Professor, 経済学部, 教授 (30186910)
YAMATO Hagime Kagoshima University, Faculty of Science, Professor, 理学部, 教授 (90041227)
OOMORI Hiroyuki The University of Tokyo, Graduate School of Economics, Associate Professor, 大学院・経済学研究科, 助教授 (60251188)
HOSHINO Nobuaki Kanazawa University, Faculty of Economics, Associate Professor, 経済学部, 助教授 (00313627)
|
|---|
| Project Period (FY) |
2002 – 2005
|
| Keywords | superpopulation model / population unique / multinomial distribution / Dirichlet distribution / local suppression |
|---|
| Research Abstract |
Throughout the years of this research our group obtained many significant results in the field of disclosure control of microdata and its applications.
Early in this research period our group edited a special volume (No.2 of Volume 51, December of 2003) of Proceedings of the Institute of Statistical Mathematics published by Institute of Statistical Mathematics on the topic of theoretical research of disclosure control techniques. This special volume contains more than 200 pages of reviews and original results of our group and many people became aware of the importance and promise of this field.
On the important problem of estimation of population uniques in a microdata set, Hoshino obtained many remarkable original results extending the Pitman model and mixed Poisson model to a new class of distributions of random partitions. Furthermore non-parametric maximum likelihood estimation by Sai became feasible and practical for actual official microdata sets after many improvements.
Furthermore as pointed out by Prof.Fienberg, who visited Japan by our invitation, we became aware of the important application of the new field of "computational algebraic statistics" to disclosure control problems. Based on this idea, Takemura established a unified strategy of disclosure risk assessment and disclosure control using decomposable models.
|
