The Supreme Court decision in the University of Michigan case in 2003 ruled the university's admissions procedures unconstitutional, giving minorities an unfair advantage of acceptance. The ruling stated race may still be used in admissions decisions to achieve diversity, but that race could not be used to give applicants preferential treatment in the admissions process. Motivated by this case, a researcher, Juan Gilbert, developed a computer based clustering method to aid admissions committees in choosing diverse entering classes. This method was evaluated using undergraduate admissions data sets from two public universities. Gilbert's method suggested diverse entering classes but did not select well based on merit. A method of improvement is introduced that maintains the academic characteristics of the university through classification, while suggesting diverse entering classes more academically similar to those actually accepted.