Superalloy FGH4095 samples were treated by laser shock processing (LSP) with laser energy of 2 J, 3 J and 4 J. Machine learning methods, including neural networks, linear regression and multitask elastic networks, were used to predict the mechanical properties (residual stress, micro-hardness and ultimate tensile strength) of superalloy FGH4095 treated by LSP. The laser energy, depth and surface micro-hardness were set as input parameters, while the residual stress, micro-hardness and ultimate tensile strength (UTS) were set as output parameters. The experimental data of untreated sample and that induced by LSP with laser energy of 2 J and 4 J were used as the training sets, and the experimental data from LSP with laser energy of 3 J were reserved as testing sets to validate the developed models. The results showed that the predicted mechanical properties obtained by machine learning methods have a good fit with experimental mechanical properties.