Photodegradation of azo dyes from industrial wastewater is challenging due to their high stability and resistance to removal. In this study, a generalized predictive model for photodegradation behavior of TiO₂ containing ZnAl layered double hydroxide (LDH) based materials in the removal process of cationic azo dyes (Rhodamine B and Methylene Blue) was proposed. The performed kinetic investigation suggested good correlation of the experimental results with theoretical settings and revealed that all photocatalysts in both photocatalytic removal reactions followed the pseudo-first order Langmuir-Hinshelwood reaction model. The inputs for artificial neural network (ANN) included four experimental variables: TiO₂ loading onto LDHs, organic dye type used for the removal process, temperature of thermal treatment of photocatalysts and reaction time, whereas for the two ANN prediction outputs removal efficiency and photo-degradation rate constants were used. The optimal topology was determined to be a three-layer feed-forward ANN with 3 input neurons and 10 hidden neurons, 3-10-1.