Statistics in Astronomy: A View Through the Looking Glass
When: Thursday, October 22, 2020, 3:00 PM - 4:00 PM
Where: Zoom event,
Title: Statistics in Astronomy: A View Through the Looking Glass
Abstract: From time to time, it is helpful to look over the shoulders of colleagues in other fields; we may be inspired to think about a problem in our own field a bit differently. It is also helpful, occasionally, to come up for air and remind ourselves of a few of the core statistical ideas that transcend fields. In this talk, I begin with the simple, but highly instructive, ON/OFF problem in astronomy in order to highlight similarities and differences between the frequentist and Bayesian viewpoints. Then, I consider the fitting of cosmological models to Type Ia supernovae data in order to illustrate the concept of parameter non-identifiability, that is, the ambiguities that can arise in fitting models to data. In view of the explosive growth of interest in machine learning, I end with a few remarks about machine learning models and algorithms.