The current investigation examined the assessment of the liquefaction potential of soil (LP) utilizing shear wave velocity (Vs) data through an evolutionary machine learning method known as multigene genetic programming (MGGP). One liquefaction index (LI) model was generated using 225 Vs data as per Andrus et al. (1999). According to the established LI model, a mathematical expression was developed for the cyclic resistance ratio (CRR) using MGGP. This expression represents the unknown limiting function, which separates the liquefaction (L) and non-liquefaction (NL) cases. The above CRR model, combined with the available mathematical formulation of cyclic stress ratio, forms the present MGGP-based deterministic approach for assessing LP in terms of factor of safety (Fs). Utilizing the above 225 Vs data, the overall rates of correct estimation for L and NL occurrences were estimated as 90.81%, 81.62%, and 68.60% by the above MGGP, existing ANN, and statistical methods, respectively, as per estimated Fs. Also, to verify the efficacy of the above evolved MGGP-based CRR formulation utilizing an independent 186 post-liquefaction Vs data as per (Juang and Chen Int. J. Numer. Anal. Meth. Geomech. 24, 1–27, 2000), the overall correct prediction rates for L and NL occurrences were estimated as 86.78%, 60.97%, and 69.39% by MGGP, existing ANN and statistical models, respectively, based on the estimated Fs.
