Statistics and Data Analysis
The aim of this 3-day course is to acquire the skills needed for analysis of experimental data and model fitting.
Objectives:
At the end of this course, a successful student will be able to:
-
Understand the basis of Bayesian analysis of data, including marginalisation.
-
Understand the principles of sampling of distributions, especially MCMC.
-
Have written an MCMC code to sample Bayesian posterior distributions, incorporating known priors on the model parameters.
-
Appreciate the need for, and be able to interpret, convergence tests.
-
Understand Bayesian Hierarchical Modelling and Simulation-based inference.
-
Be able to simulate (parts of) an experiment or model, in order to test analysis code.
Course Structure:
Each day will comprise introductory lectures in the morning (10:00-11:00, 11:30-12:30) and practical exercises in the afternoons (13:30-17:30), inferring parameters from one example from gravitational physics, Higgs boson data, or supernova data.
Day 1 - Introduction to Bayes theorem for inference, priors, sampling distributions, likelihoods and posterior probabilities. Marginalisation.
Day 2 - Introduction to sampling; Monte Carlo Markov Chain (MCMC) principles; detailed balance; Metropolis-Hastings algorithm; Hamiltonian Monte Carlo.
Day 3 - Bayesian Hierarchical Models; latent variables; Simulation-based inference.
Prerequisites:
Students should have previous familiarity with basic probability and be reasonably competent in Python scripting.
Downloads:
-
Relevant programming language - most likely python