ONERA recently computed a CFD database containing 468 Reynolds-Averaged Navier-Stokes simulations (using the Spalart-Allmaras turbulence model) about the Common Research Model wing-body-pylon-nacelle configuration. The flow conditions vary the Mach number (including transonic regimes), the angle of attack (involving flow separation), and the Reynolds number [1]. In a second step, the series of simulations has been split into a training set and a test set (respectivly 2/3 and 1/3 of the complete set), assuming that one third of the costly CFD computations should be saved and replaced by the outputs of an accurate regressor. In a first study [2] seven classical regressors of both types, global and pointwise, have been tested for the prediction of the pressure and viscous stresses at the wall of the aircraft. The conclusion were the following: (a) a large global Multi-Layer Perceptron (MLP) provided the most accurate surface fields (closely followed by IsoMap+Interpolation, Decision Tree and pointwise MLP); (b) although the R^2 values were high, the accuracy of the predicted distributions of stresses was disappointing in non-linear regions, in particular in the vicinity of shock-waves. These results have been recently completed by those provided by two additional regressors, Diffusion Map and Local Linear Embedding. Besides, a close examination of Cp discrepancy between exact and best regressors distribution in the transonic regime highlights shifts in shockwave(s) position. Machine learning algorithms with special emphasis on accuracy in strong gradient areas [3] or natural ability to represent strong gradients [4] are currently considered and the result of at least one specific method will be presented. [1] Peter J., Heib S., Hantrais-Gervois J.-L., Bennehard Q., Moëns F.. CFD database for the Common Research Model (CRM) for machine learning activities. AERO 2025, 3AF, Strasbourg France. 2025 [2] Peter J., Quentin B, Heib S., Hantrais-Gervois J.-L., Moëns F. ONERA’s CRM WBPN database for machine learning activities, related regression challenge and first results, Computers & Fluids, Vol. 302, 106838 pp. 1-11, 2025 [3] Bhaduri A., Ghosh S., Wang L., Mavris D.N. Shock wave prediction in transonic flow fields using domain-informed probabilistic deep learning. Physics of Fluids 36, 016121. 2024 [4] Catalani G., Agarwal S., Bertrand X., Tost F., Bauerheim M., Morlier J. Neural fields for rapid aircraft aerodynamics simulations. Scientific Reports 14:25496. 202