Dipartimento di Statistica Giuseppe Parenti, Universita' di Firenze
Florence, November 10-13, 1997
An introduction to graphical models
Joe Whittaker, Lancaster University
Introduction, the maths marks, the inverse variance, relationship to regression analysis
Independence and conditional independence, interaction, cross-product ratio, axioms and properties. Graph theoretic notions and independence graphs. The Markov property and separation.
Directed graphs, immoral configurations and Markov properties, chain graphs.
Graphical Gaussian models. Correlation and zero partial correlation. The inverse variance.
Model fitting by maximum likelihood. Direct and iterated estimates. Decomposable models. The deviance for testing for independence and conditional independence.
Categorical data, contingency tables.
Dipartimento di Statistica Giuseppe Parenti
Universita' di Firenze
viale Morgagni, 59
I-50134 Firenze, Italy
Lectures and practicals: 10am-2pm
Information: Giovanni M. Marchetti
Phone 055 4237211