I am currently working on a PSO which works fine and returns the cost function (which has different variables) with the lowest error rate just as expected.
Now here is my question:
Is there a way to calculate the importance of the variables of the cost function in regard to the error rate?
I want to know which variables have the biggest impact and/or if some variables have some dependencies. I know this is very similar to feature selection but i dont have a class space.