The question is in my view not well posed : an optimization technique is meant to solve a mathematical optimization problem, with objective function and constraints, defined explicitly or implicitly, Just have a look at a book about optimization theory , and, there are never cheap answers. So, for instance linear programming works on a mathematical model and not on a large dataset, unless the dataset will be transformed in an optimization problem.