Statistics course - Applied Bayesian modelling for ecologists and epidemiologists (UK)

This course is being delivered by Dr. Matt Derwood and Prof. Jason
Matthiopouus.

This extensive 6 day course will be held at SCENE (Scottish Centre for Ecology and the Natural Environment), Glasgow, United Kingdom from 26th - 31dt October 2015.

Course Aims:  This application-driven course will provide a founding in the basic theory and practice of Bayesian statistics, with a focus on MCMC modeling for ecological and epidemiological problems. Starting from a refresher on probability and likelihood, the course will take students all the way to cutting-edge applications such as state-space population modeling and spatial point-process modeling. Most importantly you should have a keen interest in ecology or epidemiology (or both) and come prepared to discuss your own research problems with the instructors.

Overview
This course provides a general introduction to Bayesian statistics,
including theory and practical implementation of MCMC methods.  By the
end of the week, you should be able to understand the key practical and
philosophical differences between Bayesian and Frequentist statistics,
have a basic understanding of how common MCMC samplers work and how to
program them, and have practical experience with the BUGS language for
common ecological and epidemiological models.  The experience gained will
be a sufficient foundation enabling you to understand current papers using
Bayesian methods, carry out simple Bayesian analyses on your own data and
springboard into more elaborate applications such as dynamical, spatial
and hierarchical modeling.  The main focus of the week is on practical
application of these methods, so a large proportion of the time will be
spent doing exercises in R.  The underlying statistical theory and an
overview of more advanced concepts will be discussed where appropriate.

Intended Learning Outcomes
By the end of this course you will be able to:
1.      Do calculations with conditional, joint and total probability.
2.      Understand the key philosophical differences between Bayesian
        and Frequentist statistics and be in a position to decide which
        approach is likely to be most useful for particular research
        questions.
3.      Use prior information along with likelihood information to form
        a Bayesian posterior in simple examples
4.      The concept of Markov chain Monte Carlo (MCMC) and how this is
        used in practice
5.      Critically discuss the role of autocorrelation and cross-
        correlation in model identifiability and Monte Carlo error
6.      Write regression models (GLMs, GLMMs) in WinBUGS / JAGS and fit
        these to data
7.      Use biological first principle or independent information to
        choose and implement both informative and minimally
        informative priors
8.      Identify when a model has converged and when sufficient Monte
        Carlo samples have been obtained
9.      Conduct model selection and comparisons using DIC.
        Understand the motivation and advantages of alternative
        model selection methods.
10.     Understand and customize more complex models for ecological
        populations in space and time Outline

Each day will consist of both taught material with discussion, and
guided computer practical sessions with assistance on hand.  These will
be interspersed evenly to ensure that all of the concepts discussed are
reinforced with practical exercises.  The planned content for each day
is as follows:

Day 1
Revision of likelihoods, using full likelihood profiles, and introduction
to the theory of Bayesian statistics.

• Probability and likelihood
• Introduction to Bayesian statistics

Day 2
An introduction to the workings of MCMC, and the potential dangers of MCMC
inference.  Participants will program their own (basic) MCMC sampler to
illustrate the concepts and fully understand the strengths and weaknesses
of the general approach.  The day will end with an introduction to the
BUGS language.

• Introduction to MCMC
• Markov chains, autocorrelation and convergence
• Introduction to BUGS and running simple models in JAGS

Day 3
This day will focus on the common models for which JAGS/BUGS would be
used in practice, with examples given for different types of model code.
All aspects of writing, running, assessing and interpreting these models
will be extensively discussed so that participants are able and confident
to run similar models on their own.  There will be a particularly heavy
focus on practical sessions during this day.  The day will finish with
a discussion of how to assess the fit of MCMC models using the Deviance
Information Criterion (DIC) and other methods.

• Using JAGS for common problems in biology
• Essential fitting tips and model selection

Day 4
The fourth day will focus on the flexibility of MCMC, and precautions
required for using MCMC to model commonly encountered datasets.
An introduction to conjugate priors and the potential benefits of
exploiting Gibbs sampling will be given. More complex types of models
such as hierarchical models, latent class models, mixture models and
state space models will be introduced and discussed.  The practical
sessions will follow on from day 3.

•     General guidance for model specification
•     State-space models

Day 5
Day 5 will give some additional practical guidance for the use of Bayesian
methods in practice, and finish with a brief overview of more advanced
Bayesian tools such as INLA and Stan.

•     Additional Bayesian methods and tools
•     Understand the usefulness of conjugate priors for robust analysis of proportions (Binomial and Multinomial data)
•     Be aware of some methods of prior elicitation
•     Strengths and weaknesses of Integrated Nested Laplace Approximation (INLA) compared to BUGS
•     Strengths and weaknesses of Stan compared to BUGS

Day 6
The final day will consist of round table discussions, the class will
be split in to smaller groups to discuss set topics/problems. This will
include participants own data where possible. After an early lunch
there will be a general question and answer time until approx. 2pm as a
whole group.

Fees:
Cost is £595 for the 6 days including lunches and refreshments or £775
for an all-inclusive option which includes the addition of accommodation,
breakfast, lunch, dinner and refreshments.  For further details or
questions or to register please email oliverhooker@prstatistics.co.uk
or visit www.prstatistics.co.uk Please feel free to distribute this
material among colleagues if you think it is suitable

Additional upcoming courses; GENETIC DATA ANALYSIS USING R; BIOINFORMATICS
FOR GENETICISTS AND BIOLOGISTS; SPATIAL ANALYSIS OF ECOLOGICAL DATA
USING R; ADVANCING IN STATISTICAL MODELLING USING R; STABLE ISOTOPE
MIXING MODELS USING SIAR, SIBER AND MIXSIAR