"Statistical Techniques for Particle Physics" (2/4)

About

This series will consist of four 1-hour lectures on statistics for particle physics. The goal will be to build up to techniques meant for dealing with problems of realistic complexity while maintaining a formal approach. I will also try to incorporate usage of common tools like ROOT, RooFit, and the newly developed RooStats framework into the lectures. The first lecture will begin with a review the basic principles of probability, some terminology, and the three main approaches towards statistical inference (Frequentist, Bayesian, and Likelihood-based). I will then outline the statistical basis for multivariate analysis techniques (the Neyman-Pearson lemma) and the motivation for machine learning algorithms. Later, I will extend simple hypothesis testing to the case in which the statistical model has one or many parameters (the Neyman Construction and the Feldman-Cousins technique). From there I will outline techniques to incorporate background uncertainties. If time allows, I will touch on the statistical challenges of searches for physics beyond the standard model and the look-elsewhere effect..

0:00:00 Slide 1

0:03:03 Slide 2

0:06:06 Slide 3

0:12:12 Slide 4

0:18:18 Slide 5

0:24:24 Slide 6

0:30:30 Slide 7

0:36:36 Slide 8

0:42:42 Slide 9

0:48:48 Slide 10

0:54:54 Slide 11