Selection of test series by a modified multidimensional mapping method

Maintaining a realistic minimum distance between compounds in a defined multidimensional parameter space ensures well-spread sets of parameter values. It has been suggested, however, that the use of this multidimensional mapping method may lead to series of compounds with high multicollinearities of parameter values. An alternative method, multidimensional mapping by distance and determinant, is discussed here. This method maximizes the determinant of the interparameter correlation matrix as well as maintaining the minimum distance criterion. Its performance is compared with other methods, and it is shown that collinearities may be overcome or maintained at low levels when this method is used.