Evaluation of machine learning approaches for estimating thermodynamic properties of new generation refrigerant R513A

•Estimation of thermodynamic properties of R513A using machine learning methods.•Determination of the most appropriate power degrees to create the polynomial estimation model.•Creating a useful polynomial estimation model with an R2 score above 0.99.•KNNRegressor gives better results with an R2 = 0.9946 for saturated liquid–vapor.•KNNRegressor gives better results with R2 = 0.9984 and above for superheated vapor.