I see that you implemented a comma strategy and i realized it was the best strategy. Also using the standard mutation and crossover strategy seems to work very well. Maybe the mutation rate could be a little bit higher.
I liked the idea you used of stagnation_limit variable as an early stopping criteria if the generation fitness is not longer improving.
Overall, seems that a different evaluation metric leads you to a better result, both in term of fitness call and fitness value. Probably with more generations you would have come to solve all the problems.
The plot are really useful to show the results!
I find your code well structured and understandable.
Hi Alessandro!
I see that you implemented a comma strategy and i realized it was the best strategy. Also using the standard mutation and crossover strategy seems to work very well. Maybe the mutation rate could be a little bit higher. I liked the idea you used of stagnation_limit variable as an early stopping criteria if the generation fitness is not longer improving.
Overall, seems that a different evaluation metric leads you to a better result, both in term of fitness call and fitness value. Probably with more generations you would have come to solve all the problems.
The plot are really useful to show the results! I find your code well structured and understandable.