A reasonable prediction of the compressive strength of lightweight high-strength concrete is an important basis for determining concrete strength. Through cluster analysis, the key factors affecting the compressive strength of lightweight high-strength concrete are found, and the degree of influence of each factor on the compressive strength is analyzed. We applied linear regression analysis of the relationship between the above factors and the compressive strength using SPSS, and performed multiple non-linear regression to establish a prediction model for the compressive strength of lightweight high-strength concrete using MATLAB. Using RMSE, we tested the simulated and actual values of the model, to determine the applicable conditions of the model through response surface analysis. The results of the study show that: cylinder compressive strength, water-binder ratio, cement dosage, coarse aggregate particle size and sand ratio are the key factors affecting the compressive strength. The R2 values of the single-factor prediction models are all greater than 0.9, and the corresponding coefficients of the lightweight high-strength concrete compressive strength prediction models are 1.46, 18.31, 21.6, –3.28, –71.12, 1.36 and 20.48 respectively; The root mean square error RMSE is all lower than 1.05 MPa. The applicable condition of the prediction model shows that the cylinder compression strength is between 3.2 MPa and 4.9 MPa. When the coarse aggregate particle size is 15 mm~25 mm, the sand ratio is 26 %~35 %, the cement dosage is 450 Kg/m³~500 Kg/m³, and the water-binder ratio is 0.34~0.4, the parameter value range is the optimal prediction space of the model. When the 5 parameters are simultaneously in the optimal prediction interval, the prediction level of the model is the best and the prediction accuracy of the proposed compressive strength prediction model is higher. The model is of great significance to the study of the mechanical properties of lightweight high-strength concrete.