Abstract:
A subclass of lattice conditional independence models
is introduced. The new class of models is called totally ordered independence models. The class is based on an assumption that the
index set which orders the random variables is a chain. It is shown
that there is a jump in the chain if and only if there is a conditional independence relation. Some comparisons between the lattice
conditional independence models and totally ordered independence
models are presented.
Key words and phrases:multivariate regression, lattice conditional independence models, totally ordered conditional independence, totally ordered multivariate linear models.