The outputs from both methods are almost optimal points, but why do people prefer optimization algorithms such as Genetic Algorithm, Fuzzy logic, Simulated Annealing etc., over MCDM techniques?
Several algorithms of artificial intelligence may be used combined to MCDA Methods as well as the fuzzy logic.
Artificial inteligence algorithms may be used to simulate the participation of a human agent (decision maker), parametrizing the MCDA methods, or analysing the results ans performing simulations with the intentio to obtain sensitivity analysis about the initial results, once to develop this kind of analysis with multiple criteria may turn very difficult for the analyst do it alone.
Fuzzy theory and logic may be used to eliminate the vagueness of the decision maker in assessing the alternatives according to each criteria.
Well, this is just ans overview about the use of these algorithms with MCDA methods, but you can find articles about this kind of approach, just for instance:
Article A credit ranking model for a parafinancial company based on ...
Conference Paper A Multi-objective Genetic Algorithm for Inferring Inter-crit...
Mathematical optimization can only be achieved by Linear Programming and perhaps with evolution techniques when you have only one objective.
When yo perform multi criteria optimization you are not looking for an optimum, but for a satisfying solution.
The reason is that it is impossible to optimize two opposite criteria, such as minimize cost and maximize benefits, it is one or the other, or a certain solution in between, that is, a satisfying solution
Actually, I would like to say that optimization is a tool that can identify what is optimized solution. However, not all the optimized solutions can satisfy the needs expectation of decision makers. This situation can be more challenging if you consider multiple decision makers with even conflicting or different set of objectives. I would like to recommend reading this paper where it discusses these situations:Article A Hybrid of Genetic Algorithm and Evidential Reasoning for O...