Department of Epidemiology & Biostatistics, Bioinformatics Presents:

Speaker: Lucy Gao, PhD, Assistant Professor of Statistics & Actuarial Science, Univ. of Waterloo

Researchers are often interested in performing an exploratory data analysis in order to generate hypotheses and then testing those hypotheses using the same data. Unfortunately, such "double dipping" can lead to highly-inflated Type 1 errors. Dr. Gao proposes a test for a difference in means between estimated clusters that solves the double dipping problem for clustering, using a selective inference framework.