Parameter estimation and all that - Part II

Least-squares method, maximum likelihood, Bayesian approach - everyone has heard about at least one of these methods. But do people understand how these approaches are related, what assumptions they make, and when to use each of them? In this talk I am going to discuss the basics of these methods and explain how they work using some simple examples. Since data analysis is the bread-and-butter of a physicist, this talk may be of interest to theorists, computer modellers and experimentalists alike.

This 2nd part of the talk will include:

• - a brief reminder how Approximate Bayesian Computation (ABC) works
• - error bars: confidence ellipse/intervals from MLM
• - Bayesian inference: credible intervals
• - fundamentals of hypothesis testing, p-value

and if time permits, Monte Carlo methods in data analysis

• - resampling methods: bootstrap, jacknife
• - Monte Carlo method of error propagation
• - Monte Carlo Markov Chain ABC
• - Monte Carlo hypothesis testing