See some answers in this post: https://www.researchgate.net/post/RMSE_or_NRMSE_Which_is_better_to_assess_numerical_model_performance_evaluation_and_why

I think by using a statistical model

The numerical results of any mathematical model can be evaluated by comparing them with related experimental data using statistical tools such as RMSE, RSS, R, R2, and others

sometimes the use of some statistical tools such as R2 does not give a clear analysis of the results, so we have to use additional tools and this topic can not be predicted until after obtaining the results and analyzing them.

see paper

https://www.pc-progress.com/Documents/Jirka/Phogat_et_al_Errig&DrainSystemEngin_2016.pdf

Thank you so much for your answers Anton Vrdoljak Kn Basim Ziad Tark Abd Ali https://www.researchgate.net/profile/Maged_Bin-Saad

Could be ANOVA.

One and two-way ANOVA....

You can use ANOVA or analysis of covariance or use minitab for the analysis

You can use mike11,21 program

Really, one can not emphasis what the criterion can use to sure from convergence in generally. But from its problem may usually draw out suitable criterion.

@Waqed Al-mussawi

Use MATLAB and see , its friendly nature .

Thank you so much for your answers Haider I. Alyasari Naseem Ahmad Mehdi Mesrizadeh Denys Dutykh

The akaike razor can ve useful to evaluate models or área validación metrix but you'll need many validación data set