I have designed a hybrid Evolutionary Algorithm for my research using Genetic Algorithms. Genetic Algorithms have different operators such as Selection, Crossover and Mutation. These operates are executed on a number of chromosomes in a population. Crossover and Mutation are executed depending on a pre-determined value. A random number is drawn and if it is higher than this pre-determined value, the operator shall be executed, else it is not.
I need to make different tests using different population sizes and different values for both Crossover and Mutation. I call each combinations of these values, settings.
Now, If I ran two different settings for 10 times using the same algorithm, and decided to use Mann-Whitney Wilcoxon U Test. My questions are:
- What does the test really measure here?
- What can I infer from the output p-value?