In the original variable space, the MD takes into account the correlation in the data, since it is calculated
using the inverse of the variance–covariance matrix of the data set of interest. However, the computation of the
variance–covariance matrix can cause problems. When the investigated data are measured over a large number
of variables e.g., NIR spectra., they can contain much redundant or correlated information. This so-called multicollinearity
in the data leads to a singular or nearly singular variance–covariance matrix that cannot be inverted.