Gene Expression Programming in Open Channel Hydraulics
While the complex relationship between variables can be represented by regression analysis, machine learning methods have proven to be more efficient in representing these complex relations. Relationship between variables is expressed in form of expression tree (ET) as developed by Ferreira. The ET represents the gene via simple linear rules, which are the basics for the unequivocal Karva language. Gene mutation is the strength of the Gene Expression Programming (GEP) allowing genes and thus expressions to continuously evolve for better ones until the best and most fit offspring expression is reached. The mutation can occur in any part of the gene head and tail. In most studies, the GEP models are developed using available experimental data and there is always a degree of uncertainty of prediction outcomes; therefore, it would be useful assess the quantitative effect of the stochastic nature of the GEP models.