In this paper, six parameters of the hull of a single hull model, namely the aspect ratio, aft body length/fore body length, width/height of break step, slope elevation at break step, forebody distortion and aft body keel Angle, were changed. And the three test parameters, static load coefficient, longitudinal position of center of gravity and speed, with a total of nine dimensions, are used as the regression prediction input. By studying the hydrodynamic performance of single hull gliding in static water under different hull form machine learning method of Gaussian process and random forest was used to select different kernel functions to construct the hydrodynamic performance database of single hull. The experimental data of single hull resistance, attitude and heave were taken as output, and MAE and R2 were taken as evaluation indexes to compare the difference between the real and predicted values. The results show that both Gaussian process and random forest have better prediction results, but the Gaussian process is better than the random forest method. Moreover, in the calculation results of selecting different Gaussian process kernel functions, the precision of selecting multi-kernel mixed kernel function is the best. Therefore, the Gaussian process machine learning method can achieve better prediction accuracy, and further save the test cost and improve the efficiency.