Tech Reports

Tech Report kmi-98-05 Abstract

Learning Conditional Probabilities from Incomplete Data: An Experimental Comparison
Techreport ID: kmi-98-05
Date: 1998
Author(s): Marco Ramoni and Paola Sebastiani

This paper reports some experimental results comparing three parametric methods, Gibbs Sampling, EM algorithm and Bound and Collapse, for the estimation of conditional probability distributions from incomplete databases. 1. Knowledge Media Institute, The Open University. 2. Department of Actuarial Science and Statistics, City University.