Iterative Scaling in Curved Exponential Families
Abstract
The paper describes a generalized iterative proportional fitting procedure which can be used for maximum likelihood estimation in a special class of the general loglinear model. The models in this class, called relational, apply to multivariate discrete sample spaces which do not necessarily have a Cartesian product structure and may not contain an overall effect. When applied to the cell probabilities, the models without the overall effect are curved exponential families and the values of the sufficient statistics are reproduced by the MLE only up to a constant of proportionality. The paper shows that Iterative Proportional Fitting, Generalized Iterative Scaling and Improved Iterative Scaling, fail to work for such models. The algorithm proposed here is based on iterated Bregman projections. As a byproduct, estimates of the multiplicative parameters are also obtained.
 Publication:

arXiv eprints
 Pub Date:
 July 2013
 arXiv:
 arXiv:1307.3282
 Bibcode:
 2013arXiv1307.3282K
 Keywords:

 Statistics  Computation;
 Statistics  Methodology;
 62J12
 EPrint:
 The paper has one figure