The accurate and real-time prediction of the mechanical property response for the safety assessment of the marine riser is necessary during offshore oil and gas production. However, a finite element simulation is time-consuming, which obstructs the real-time monitoring and evaluation of the marine riser. Therefore, to obtain the real-time mechanical response of the marine riser, a predicting method based on the simulation and data-driven models is proposed. To obtain the prediction data, the finite element simulation model under the typical working conditions is established by the Abaqus software, and then the sensitivity analysis of influence factors is conducted due to the varied working conditions. The orthogonal experimental method is used to select the typical working conditions of the marine riser, and the selection of the sensitive influence factors is ensured based on the typical working conditions, the range analysis, and the grey relational method. To obtain enough training data, the Latin hypercube sampling method is used to enlarge the data for the analysis process. To find a more reliable prediction algorithm, four typical single surrogate models, which are the polynomial response surface model, the radial basis model, the kriging model, and the support vector machine regression model, are selected to establish a predicting model library. Meanwhile, a combining surrogate model, including the average weight surrogate model and the adaptive weight surrogate model, is established based on the combining weight to further improve the predicting accuracy. As a result, an optimal surrogate model is selected from the above four single and two combining surrogate models based on four evaluation criteria including the mean absolute percentage error, the mean absolute error, the root mean square error, and the goodness of fit. The cases show that the optimal surrogate model has a highly rapid response, with average relative errors are 8.17%, 3.42%, and 11.40% for the maximum bending moment, the maximum equivalent stress, and the maximum lateral deformation, respectively. In the above study, the accurate and reliable prediction method of the mechanical property responses will provide a theoretical basis for ensuring the safety of the marine riser.