Abstract: 120 tests of metakaolin-slag geopolymer paste were designed, and the effects the alkali activator concentration, modulus and liquid-solid ratio on the compressive strength, fluidity and setting time of geopolymer paste were discussed. Based on the obtained experimental data, a Lasso multiple regression model was established to predict the 7 days and 28 days compressive strength, fluidity and initial and final setting time of metakaolin-slag geopolymer paste. Experimental results show that: (1) The compressive strength of geopolymer increases with the increase of the concentration of alkaline activator, decreases with the increase of the liquid-solid ratio, and first increases and then decreases with the increase of the modulus. (2) With the increase of liquid-solid ratio, the setting time is prolonged. The influence of the modulus and concentration on the setting time of geopolymer is determined by the silicon content and alkali content of the alkaline activator. (3) The fluidity is mainly related to the viscosity and the liquid-solid ratio of the alkaline activator. The verification results of the model show that: The Lasso algorithm is used to regularize the regression model, which avoids the overfitting phenomenon caused by excessive regression coefficient. The proposed regression model can accurately predict the macroscopic properties of metakaolin-slag geopolymer paste, and the correlation coefficient between the predicted value and the test value in the test set data is greater than 0.92.