A branch search algorithm for maximum likelihood paired comparison ranking
John A. Flueck, James F. Korsh · Biometrika · 1974
An efficient algorithm is presented for solving a general class of mathematical programming problems. This class includes the specific task of finding a maximum likelihood ranking based on paired comparison data. It differs from a branch and bound scheme in that it searches the nodes of a tree one branch at a time. As a result, its time requirements appear to be better and its storage requirements are better than or as good as previous algorithms. An example from the literature is presented for illustration and comparison of this method with previous work. Computational experience is described for an important class of problems.