: Seragged joint is a common structural surface form in rock mass, its special geometric form and structural characteristics have an important influence on the shear strength and stability of rock mass. In order to overcome the problem that it takes a lot of manpower and material resources to obtain the shear strength of segmented rock mass in in situ test, the application of group intelligent optimization technology and interpretable data driven method in the shear strength prediction of zigzag segmented rock mass is explored. First, 50 sets of zigzag rock sample data were collected to form the sample database, including the basic friction Angle of rock, normal stress of joint surface, ratio of normal stress to tensile strength of complete rock, joint inclination angle and shear strength. Subsequently, four machine learning-based data-driven models and two combined models based on group intelligent optimization algorithms were constructed to predict the shear strength of zigzag segmented rock mass, directly establishing the nonlinear relationship between input parameters and shear strength. Finally, the prediction performance of each model was quantified by using the coefficient of determination (R2), corrected coefficient of determination (Adj.R2), root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and variance contribution rate (VAF). The results show that CSA-XGBoost is the best prediction model, R2(0.980), Adj.R2 (0.964), RMSE (0.335), MAE (0.265), MAPE (5.72%), VAF (99.1%), which can effectively predict the shear strength of zigzag rock mass. In addition, TreeSHAP method is used to quantify the degree of influence of different characteristics on the shear strength of zigzag rock mass, and found that the normal stress and inclination Angle have a great influence on shear strength, which needs to be paid attention to and considered in engineering design and construction.