Quantitative structure–affinity relationship study of azo dyes for cellulose fibers by multiple linear regression and artificial neural network

• A structure-affinity study of the azo dye affinity on cellulose fibers was performed.
• Robust and predictive power models were obtained by MLR and ANN.
• Increased hydrogen-bonding donor ability increased the dye affinity.
• Group polarizability and electronegativity were significant for cellulose binding.

A quantitative structure–property relationship study was performed to correlate descriptors representing the molecular structures to fiber affinities for azo dyes. The complete set of 51 compounds was divided into a training set of 41 compounds and a test set of 10 compounds by using DUPLEX algorithm. Multiple linear regression analysis was used to select the best subset of descriptors and to build linear models; nonlinear models were developed by means of artificial neural network. The robustness of the obtained models was assessed by different approaches, including leave-many-out cross-validation, Y-randomization test, and external validation through test set. The obtained models with four descriptors show good predictive power: for the test set, a squared correlation coefficient (r2) of 0.916 was achieved by the linear model; while the nonlinear model with r2 of 0.935 for the test set performs better than the linear model. Furthermore, the applicability domain of the models was analyzed based on the Williams plot. The donor atom number for H-bonds, group polarizability and electronegativity of the dye molecules are found to play important roles for the dye-fiber affinity.