NiTi SMAs, also known as Nitinol, are well-known and widely used due to their unique properties. This study predicts the transformation behaviour of a binary near-equiatomic shape memory alloy (SMA) during thermal cycling using empirical and ANN-based models. The input data was generated through thermal cycling tests using a differential scanning calorimeter (DSC) under a nitrogen atmosphere, wherein the maximum and minimum temperatures were varied based on the transformation temperatures of the alloy. Three different models, i.e. symmetrical, asymmetrical and artificial neural network (ANN), were developed to understand the transformation behaviour of the alloy using the same set of test data for validation. For qualitative and quantitative comparisons of the model, priority was given to the simplicity of the model (minimum variables) and the accuracy of the prediction. The results show that the ANN-based model can predict the transformation behaviour more accurately (99.81%) as compared to the conventional empirical models, i.e., symmetric (96.64%) and asymmetric models (98.14%).