At high-redshift CIV EQW and blueshift capture most of the diversity of the broad emission-line properties of quasars. It has been posited that the relationships between these two properties suggest a two-component (disk-wind) model of the broad-emission-line region (BELR) where the component that dominates is a function of L/LEdd. In order to correctly classify quasars and further investigate physical models we must first know how accurately single-epoch spectroscopy can locate quasars in emission line parameter space when multi-epoch spectroscopy is not available. We aim to ascertain the true distribution of quasars in CIV parameter space and how variability (as well as noise) may result in the movement of an individual object in the parameter space over time. We investigate the diversity of SDSSRM quasars using independent component analysis (ICA); a computational technique which is able to reconstruct spectra in a form similar to PCA (Principal Component Analysis), but which does not require orthogonality in its components. The weights produced by ICA contain information regarding the correlations between spectral features, which therefore results in an increase in the S/N of the spectra. The subset of the SDSSRM sample we use consists of 133 quasars observed up to 53 times over a period of 3 years, providing a large sample of data and different baselines of variability.