def get_result(self) -> R:
ranking = FastNonDominatedRanking(self.dominance_comparator)
ranking.compute_ranking(self.solutions)
try:
non_dominated = ranking.get_nondominated()
except IndexError:
non_dominated = None
return non_dominated if non_dominated else self.solutions
no_evaluations = Paras.N_GEN*Paras.N_IND
algorithm = NSGAII_MEDA(
problem=problem,
population_size=Paras.N_IND,
offspring_population_size=Paras.N_IND,
mutation=IntegerPolynomialMutation(probability=1.0 / problem.number_of_variables, distribution_index=20),
crossover=IntegerSBXCrossover(probability=0.8, distribution_index=20),
termination_criterion=StoppingByEvaluations(max_evaluations=no_evaluations)
)
progress_bar = ProgressBarObserver(max=no_evaluations)
algorithm.observable.register(progress_bar)
algorithm.run()
solutions = algorithm.get_result()
non_dominated_ranking = FastNonDominatedRanking(algorithm.dominance_comparator)
non_dominated_ranking.compute_ranking(solutions)
non_dominated_sols = non_dominated_ranking.get_nondominated()
plot_front = Plot(title='Pareto front approximation', axis_labels=['Dis', 'Mani'])
plot_front.plot(non_dominated_sols, label='NSGAII-ZDT1')
list_labels = []
for sol in non_dominated_sols:
list_labels.append(sol.variables)
dis = sol.objectives[0]
man = sol.objectives[1]
# train_acc = sol.objectives[2]
Yt_sol = sol.variables
acc = (np.sum(Yt_sol == problem.Yt)+0.0)/len(Yt_sol)
print("Dis: %.5f, man: %.5f, acc: %.5f" %(dis, man, acc))
# print("Dis: %.5f, man: %.5f, train acc: %.5f, acc: %.5f" % (dis, man, train_acc ,acc))
vote_label = Helpers.voting(list_labels)
