Application of principal component-wavelet neural network in spectrophotometric determination of acidity constants of 4-(2-thiazolylazo)-resorcinol

► WNN as a powerful tool for the prediction of acidity constant of TAR using UV-Vis spectra. ► Spectra were generated by distribution function, mass balance law and Gaussian curves. ► Model is able to predict pKa of various diprotic acids using only UV-Vis spectra. ► The small relative errors for predicted pKa show applicability of ANN and WNN models. ► In comparison with ANN, WNN converges faster and has slightly better accuracy.