In this paper we propose two new descriptive measures for multivariate data: the effective variance and the effective dependence. These measures have a direct geometric and statistical interpretation and can be used to compare groups with different number of variables. The contribution of these...

Combining multiple forecasts provides gains in prediction accuracy. Therefore, with the aim of finding an optimal weighting scheme, several combination techniques have been proposed in the forecasting literature. In this paper we propose the use of sparse partial least squares (SPLS) as a method...