Y. Migron et al., COMPUTER-SEARCH FOR SOLVENT AND SOLUTE PARAMETERS THAT DETERMINE PARTITION DATA, Chemometrics and intelligent laboratory systems, 22(2), 1994, pp. 191-197
Solute partitioning between water-immiscible organic solvents and wate
r can be summarized by the matrix product D = PCQ(T). Here P is the ma
trix of solute properties, Q is that of solvent properties (Q(T) is it
s transpose), and C describes the weight of each contribution to the d
ata (log partition coefficient) matrix D. Factor analysis, employing s
ingular value decomposition, yields an 'improved' data matrix DBAR and
an 'improved' matrix CBAR. Factor analysis and target factor analysis
, applied to the 28 x 9 matrix D, showed that there are three principa
l factors and allowed their selection from among the various parameter
s offered for this analysis.