GP has been establishing to display a few focal points over other data-driven models (DDMs). Its significant favorable position is in its capability to create programs that can proficiently simulate complex procedures utilizing symbolic expressions [75]. Another point of preference of GP over other robust methods, for example, SVM is that it creates a straightforward and organized representation of the framework being demonstrated, without requiring from the earlier recognizable of the model structure. Be that as it may, in GP both the model structure and its parameters are being optimized, as they are both part of the search process. This gives GP the capacity to naturally recognize the data variables that contribute advantageously to the model and ignore those that don't, in this way reducing the dimensionality of the model. Additionally, GP advances models equipped for giving physical understanding into the input-output interactions inherent in demonstrated framework, rather than the SVM where difficulty still exists in extracting knowledge from the parameters. Then again, GP has its own particular restrictions. Basically, GP is not very powerful in discovering constants, and more importantly, it tends to create more complex functions as the forecast horizon growths [76].