Date: 20th - 21st July 2011
Duration: 2 Days
Location: Postgradute Statistice Centre, Lancaster University.
Presenter: Dr Joe Whittaker
Cost
* External from a non academic institution - £440
* External from an academic institution - Staff - £120
* External from a academic institution - Postgraduate Student - £60
Course fees include all supporting documentation, refreshments and lunches.
To book and for further information please visit http://shortcourses.maths.lancs.ac.uk/graphicalmodels
Course Description
A particular problem of statistical data analysis is to model the
inter-dependencies among a set of response variables. Graphical
models, based on conditional independence, provide a powerful and
informative toolbox for unravelling manifest interactions and
associations.
This course discusses undirected graphical models, describing their
properties and their statistical analysis. The course is not
mathematical, but sometimes phrased in mathematical notation, and is
loosely based on material from Whittaker (1990) Graphical models in
applied multivariate statistics, Wiley.
Novel features of this course are the emphasis on weighted
independence graphs, a review of modern search methods and a survey of
R packages useful to graphical modelling. Many examples of successful
analyses drawn from the social, biological and physical sciences are
given.
The course is aimed at researchers and research students who have
experience of statistical modelling (up to linear regression) and
hypothesis testing, who wish to develop techniques to analyse
multivariate data.
The aim of the course is to provide a background of theory with
opportunities to apply the techniques in practice, and each session
consists of a lecture/ demonstration and a practical. Participants
are encouraged to bring their own data set to explore during the
course. The software package used is R and participants are expected
to have some basic familiarity. Using R from inside a statistical
package such as SPSS is also briefly outlined.
The following topics are covered
* conditional independence, independence graphs, mutual information.
* graphical Gaussian models, maximum likelihood estimates, explained
information.
* graphical log-linear models for categorical data, MLEs,
* model search, trees, score and constrained search.
The illustrations use R-packages such as gRbase, Rgraphviz, ggm, minet
and several others.
Learning
Participants learn through the application of concepts and techniques
covered in the course to real data sets, including their own. Students
are encouraged to examine issues of substantive interest in these
studies. Successful students will be:
* familiar with graphical models and conditional independence
* able to investigate data with exploratory graphical models
* able to confirm hypotheses encapsulated in models
* able to apply theoretical concepts
* able to identify and solve problems
* able to analyse data using appropriate techniques and interpret statistical
output
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|