We investigate the possibility of exploiting partial correlation graphs for identifying interpretable latent variables underlying a multivariate time series. It is shown how the collapsibility and separation properties of partial correlation graphs can be used to understand the relation between...

In this paper a projection pursuit method is developed which determines optimal multivariate latent factor models based on a flexible loss function. This way, the unknown model coefficients are estimated with respect to optimal predictive power. The specification of the loss function in...