Machine learning models for fast selection of amino acids as green thermodynamic inhibitors for natural gas hydrate

•Machine learning to predict gas hydrate formation temperature (HFT) with amino acid.•Relative deviation of predicted HFT was less than 0.2% with experimental HFT.•Coefficient of determination (R2) of the support vector machine model was 0.998.•HFT was much more sensitive to the concentration than the type of amino acids.