Dr Robin Evans
Tutorial Fellow in Statistics, Jesus College
robin.evans at stats.ox.ac.uk
+44(0)1865 272860 (Department)
+44(0)1865 282859 (Direct)
+44(0)1865 272595 (Fax)
Graphical models, causal inference, latent variable models, algebraic statistics
'All the world’s a DAG, and all the men and women merely vertices.'
My postdoctoral research at Cambridge was mainly concerned with understanding the marginal distributions of directed acyclic graph (DAG or Bayesian network) models, and how that might be applied to causal inference. Many multivariate random systems around us can be well modelled as a DAG, but in which we are only able to observe some of the random quantities; we wish to be able to recover the hidden structure from what we do observe, wherever this is possible.
This work has led in two directions: firstly, showing that the algebraic constraints already known are complete, and secondly finding good ways of deriving additional semialgebraic constraints (i.e. inequalities). Such constraints can be used to (partially) identify causal quantities of interest. I am also interested in the mathematical and statistical properties of more general latent variable models, conditional independence, and model parametrizations.
I received my PhD in Statistics from the University of Washington in 2011, and was a Postdoctoral Research Fellow at the Statistical Laboratory in Cambridge from 2011 to 2013.
R.J. Evans and T.S. Richardson, Marginal loglinear parameters for graphical Markov models. Journal
of the Royal Statistical Society, Series B, 75 (4) pp 743-768, 2013.
I. Shpitser, R.J. Evans, T.S. Richardson and J.M. Robins, Sparse nested Markov models with loglinear
parameters. In Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence (UAI13),
pp 576-585, 2013.
R.J. Evans, Graphical methods for inequality constraints in marginalized DAGs. In Proceedings of the
22nd Workshop on Machine Learning for Signal Processing, 2012.
T.S. Richardson, R.J. Evans and J.M. Robins, Transparent parametrizations of models for potential
outcomes (with discussion). Bayesian Statistics, 9, pp 569-610, 2011.