CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Split sample hypothesis tests and confidence sets for mixture models Authors:  Hien Nguyen - University of Queensland (Australia) [presenting]
Abstract: Recent work has demonstrated that one can construct hypothesis tests and confidence regions for arbitrary statistical models in a finite sample correct and estimator agnostic manner, using a split data likelihood ratio methodology. We demonstrate some interesting and useful applications of these methods in the mixture model context.