Collect. Czech. Chem. Commun. 1996, 61, 713-725

Principal Component Regression with Chemical Shift Increments. I. p-Disubstituted Benzenes and 2-Naphthyl Derivatives

Miroslav Holík

Department of Theoretical and Physical Chemistry, Masaryk University, 611 37 Brno, Czech Republic


Prediction of 13C substituent chemical shifts in 14 series of para-disubstituted benzenes and in 2-substituted naphthalenes was based on principal component regression with chemical shift increments for the ipso, ortho, meta and para position of monosubstituted benzenes. Mean-centered matrix of shift increments was submitted to singular value decomposition and principal component regression was used for the projection of the investigated substituent chemical shifts and for the calculation of regression coefficients. Residual standard deviation between experimental and fitted values in para-disubstituted benzenes was in agreement with absolute values of "an electron demand" of substituents. Inspection of the regression parameters revealed that for the prediction of chemical shifts in 2-substituted naphthalenes the combination of chemical shift increments was better than the use of single increments. It is believed that the presented procedure is general and can be used for other aromatic or heteroaromatic systems.

Keywords: Linear transformation; Multivariable regression analysis; Principal component regression; Chemical shift increments; 13C NMR substituent chemical shifts.