MSc in Applied Statistics: MCMC module MT 2015

Course details

Eight lectures in the SPR 1 lecture theatre and two 1.5-hour practicals in the SPR1 Labs.

 

Sample solutions to the assessed practical from Week 8 MT15 are available here and below.

 

Lecture material

Lectures 1-2: concepts and motivation; inversion; transformation; rejection (related R-code L1-4).

Lectures 3-4: Rejection (continued); Importance sampling and variance reduction;

Lectures 5-6: Markov chains; Metropolis Hastings MCMC (related R-code L5-6).

Lectures 7-8:  Convergence and Variance; MCMC flavors; R-JAGS (R-code and Bugs directory L7-8).

 

Prof Ripley’s lecture notes from 2013 are recommended reading for lecture 8, which is based on these notes.

 

Problem classes

I have arranged classes in week 6 and 8, Fridays 11:30-12, 12-12:30 and 1-1:30

Problem sheets on Rejection and Importance Sampling and MCMC

Sample solutions to problem sheets on Rejection and Importance Sampling and MCMC

 

Practical classes

Week 6: non-assessed practical and sample solutions for Rejection sampling and Bayesian inference.

Week 8: assessed practical (due 12 Midday Monday week 9 MT15). AustFirstAids.txt data file,

GammaMixtureData.txt, and sample solutions to the second exercise and sample solutions to the

third assessed exercise. My solution is just one of many ways to complete the exercise satisfactorily.

Here is a temporary link to the jags user manual for the first exercise in the week 8 practical.

Reading

 

Recommended reading

C.P. Robert and G Casella, ``Introducing Monte Carlo Methods with R'

 

Advanced texts

W. Venables and B.D. Ripley, “Modern Applied Statistics with S”, ISBN 0- 387-95457-0

 

Quotes

 

"Anyone who can do solid statistical programming will never miss a meal."

(Prof David Banks 2008)

 

"I keep saying that the sexy job in the next 10 years will be statisticians."

(Hal Varian, Chief Economist at Google)

 

"Hal was right, but they didn’t call it 'Statistics'." (Prof David Hand, 2012,

referring to the many names – Data Assimilation, Quantitative Analysis, Big Data,

Data Analysis, Data Science - we have for 'Statistics')

 

 

 

Geoff Nicholls

nicholls@stats.ox.ac.uk