Lab 3 review by Giorgio Cacopardi #309685

[GioC1810]

Hi Roberto!
Firstly, good job on the lab 3, your code was very clear and well structured.
I really appreciate the division of the various section and the report of your different approaches, it shows your improvements step by step and that especially in this kind of problem, it is really important to understand what improvement make to obtain better results
However, i have some suggestion that i hope can help you:

• About your Individual class, i suggest to define an "init" method and define directly there the genome random generation and the fitness calculus, to make lighter the code of the evolutionary algorithm and avoid cycle to calculate the fitness for new Individuals
• I think that check the average and eventually increase your mutation rate is very clever, i would suggest also to add a control on the standard deviation to check if the algorithm get stuck in some local optimal and eventually remove some individuals by using for instance the stochastic universal sampling , and then randomly insert new individual to reach again the population size
• You can try to check if mixing crossover and mutation at the same time can lead to better result, like for example use to generate a child not only the mutation or the recombination or the mutation but both of them simultaneously

Hoping that my advices can help,
Keep going!