TY - JOUR

T1 - Fibers of multi-way contingency tables given conditionals

T2 - relation to marginals, cell bounds and Markov bases

AU - Slavković, Aleksandra

AU - Zhu, Xiaotian

AU - Petrović, Sonja

N1 - Funding Information:
A. Slavković and X. Zhu supported in part by NSF grants SES-052407 and BCS-0941553 from the Pennsylvania State University. S. Petrović supported in part by grant FA9550-12-1-0392 from the U.S. Air Force Office of Scientific Research (AFOSR) and the Defense Advanced Research Projects Agency (DARPA). The authors would like to thank the reviewers for their valuable insights and comments.
Publisher Copyright:
© 2014, The Institute of Statistical Mathematics, Tokyo.

PY - 2015/8/22

Y1 - 2015/8/22

N2 - A fiber of a contingency table is the space of all realizations of the table under a given set of constraints such as marginal totals. Understanding the geometry of this space is a key problem in algebraic statistics, important for conducting exact conditional inference, calculating cell bounds, imputing missing cell values, and assessing the risk of disclosure of sensitive information. Motivated by disclosure problems, in this paper we study the space of all possible tables for a given sample size and set of observed conditional frequencies. We show that this space can be decomposed according to different possible marginals, which, in turn, are encoded by the solution set of a linear Diophantine equation. Our decomposition has two important consequences: (1) we derive new cell bounds, some including connections to directed acyclic graphs, and (2) we describe a structure for the Markov bases for the given space that leads to a simplified calculation of Markov bases in this particular setting.

AB - A fiber of a contingency table is the space of all realizations of the table under a given set of constraints such as marginal totals. Understanding the geometry of this space is a key problem in algebraic statistics, important for conducting exact conditional inference, calculating cell bounds, imputing missing cell values, and assessing the risk of disclosure of sensitive information. Motivated by disclosure problems, in this paper we study the space of all possible tables for a given sample size and set of observed conditional frequencies. We show that this space can be decomposed according to different possible marginals, which, in turn, are encoded by the solution set of a linear Diophantine equation. Our decomposition has two important consequences: (1) we derive new cell bounds, some including connections to directed acyclic graphs, and (2) we describe a structure for the Markov bases for the given space that leads to a simplified calculation of Markov bases in this particular setting.

UR - http://www.scopus.com/inward/record.url?scp=84931566233&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84931566233&partnerID=8YFLogxK

U2 - 10.1007/s10463-014-0471-z

DO - 10.1007/s10463-014-0471-z

M3 - Article

AN - SCOPUS:84931566233

SN - 0020-3157

VL - 67

SP - 621

EP - 648

JO - Annals of the Institute of Statistical Mathematics

JF - Annals of the Institute of Statistical Mathematics

IS - 4

ER -