Most of the existing evaluation methods for the groutability of sand stratum are classification evaluation methods, and the classification standards are different, which is not conducive to practical engineering application. To solve this problem, particle swarm optimization (PSO) was used to optimize the least squares support vector machine (LSSVM), and a quantitative evaluation model of sand groutability was proposed. With water/cement ratio (R_WC), relative density D_r, fine sand content θ (diameter<0.075 mm), and sand characteristic particle size D₁₀ and D₁₅ as control variables, 129 groups of groutability tests were conducted, and grout diffusion distances of each group were measured and used as quantitative evaluation indexes. The relational model between groutability factors and the control variables was obtained by PSO-LSSVM method. Furthermore, the global sensitivity of groutability factors was analyzed by FAST method. Results show that the grout diffusion distance predicted by the PSO-LSSVM model in the test set was close to the test value, the goodness of fit R² was 0.982, and the prediction model had a high prediction accuracy. The sensitivity sequence of groutability factors was: D₁₀>D₁₅>D_r>θ>R_WC, where D₁₀ and D₁₅ were significantly more sensitive than D_r, θ, and R_WC.