Mr. Brian Williamson, a doctoral candidate in biostatistics at the University of Washington School of Public Health, was among three students this year to win Student Paper Awards from the Nonparametric Statistics Section of the American Statistical Association. The winning papers were presented at the 2019 Joint Statistical Meetings in Denver.
Mr. Williamson’s paper, “A unified approach to nonparametric variable importance assessment,” proposes a model-agnostic measure of variable importance. This allows estimates of importance to be compared across different machine learning algorithms. The work develops a rigorous framework for estimating and performing valid statistical inference on the true importance.
While the procedure is broadly applicable, one area in which it has been employed is the study of vaccine efficacy. The proposed approach allows investigators to use state-of-the-art machine learning tools to estimate important variables for predicting when a vaccine is effective. This, in turn, may help to more quickly develop new candidate vaccines for a range of diseases.Friday Letter Submission, Publish on September 13