Title: On maximum likelihood estimation for mean zero versus general Ising graphical Markov models
Authors: Giovanni Maria Marchetti - University of Florence (Italy) [presenting]
Nanny Wermuth - Chalmers University of Technology (Sweden)
Abstract: The properties of the class of mean zero Ising models in fitting graphical Markov models to binary data are summarized. Moreover, we address parameter estimation by maximum likelihood with a comparison to the larger class of general Ising models. We discuss how to simplify estimation using mean zero Ising models when the marginal distributions of data have skewed margins.