Introduction to graphical models and Bayesian networks for social scientists

Roger Bannister Lecture Theatre

Faculty of Medicine

Imperial College London

St. Mary's Campus, London

15-16 October 2009

Lecturers

Dr Sara Geneletti of the London School of Economics Statistics Department and Dr Gianluca Baio of University College London department of Epidemiology and Public Health.

Course outline

This 1.5 day course has two components. Day 1 covers the basics of graphical models, also known as causal diagrams, focusing on directed acyclic graphs. Day 2 is a half-day introduction to Bayesian networks and expert systems with Hugin software which used graphical models to inform decision making processes.

Day 1: Diagrams are often used to express belief about the relationships between variables in complex problems. By formulating such diagrams as graphical models and in particular directed acyclic graphs (DAGs) it is possible to formalise these diagrams and use them for statistical and causal inference. Also, by representing complex problems using DAGs it is possible to build a bridge between researchers in different fields.

Day 2: We cover the basics of Bayesian inference, with specific reference to social sciences and its relationships and possible applications with the technology of Probabilistic Expert Systems (Bayesian Networks). The main general issues are discussed in a tutorial form with worked examples in Hugin, one of the main computer softwares used in the field.

This course is aimed at introducing graphical modelling concepts to social scientists with a basic understanding of statistical methods (e.g. hypothesis testing, regression).

Tentative schedule:

Day 1:
- 10:30-11:00 Registration + Coffee
- 11:00-12:30 Lecture: Basic definitions
- 12:30-13:15 Lunch
- 13:15-14:45 Lecture: Drawing DAGs
- 14:45-15:00 Coffee break
- 15:00-16:15 Lecture: Confounding and other bias
- 16:15-17:00 Lecture: Overview of software and discussion
Day 2:
- 9.30-10.30 Lecture: Introduction to Bayesian inference and the use of Probabilistic Expert Systems
- 10.30-11.00 Coffee Break
- 11.00-12.00 Exercise/Tutorial with examples in Hugin
- 12.00-12.30 Conclusions/Discussion

The course material is now available for download from here

Location

The course will take place in the Division of Epidemiology, Public Health and Primary Care of the Faculty of Medicine and the participants will be able to use these facilities. Note that this is in the Faculty of Medicine, Imperial College London at St. Mary's Campus, and NOT in the main campus in South Kensington.

The complete address is

Faculty of Medicine
Imperial College London
St. Mary's Campus, Norfolk Place
W2 1PG London - UK

How to get to the Department of Epidemiology & Public Health

http://www.streetmap.co.uk/map.srf?x=526841&y=181421&z=106&sv=526841,181421&st=4&ar=y&mapp=map.srf&searchp=ids.srf&dn=860&ax=526841&ay=181421&lm=0

More information on how to arrive can be found at http://www1.imperial.ac.uk/medicine/contacts/campuses/stmarys/

By Car

St Mary's is situated outside of the congestion charge zone but please note that there are no College parking spaces available.

By Underground/Rail

Paddington tube stations (Bakerloo, District, Circle and Hammersmith/City line) are two minutes walk from St Marys Campus Paddington Railway Station is also a two minute walk away.

By Bus

Buses 7, 15, 23, 27 and 36 stop directly outside St Mary's campus.

Booking and course fees

£25 per day for postgraduate students registered at UK academic institutions
£50 per day for staff at UK academic institutions, ESRC funded researchers and registered charity organisations
£195 per day for all other participants

This includes the course materials, refreshment and lunch.

In order to book a place for the course, you should contact Sara Geneletti (s.geneletti@imperial.ac.uk).

https://www5.imperial.ac.uk/medicine/coursebookings/intrographmodbayesiannetwork_0910.html

If you have any queries, contact Madeleine Kirk (m.kirk@imperial.ac.uk)