label='Mean')
axes[1, 1].legend()
#plt.figure(figsize=(8,6))
for i, n_p in enumerate(n_pop):
axes[2, 1].errorbar(range(o),
mean_time[i, :],
std_time[i, :],
label='Pop size = ' + str(n_p))
axes[2, 1].set_xticks(range(o))
axes[2, 1].set_xticklabels(n_offspring)
axes[2, 1].errorbar(range(o),
np.mean(mean_time, axis=0),
np.mean(std_time, axis=0),
linestyle='--',
linewidth=2,
label='Mean')
axes[2, 1].legend()
KP_instance = KPSolver("instances/kp/kp-n279.txt")
TSP_instance = TSPSolver("instances/tsp/tsp-a280.txt")
TTP_inst = TTPInstance("instances/ttp/ttp-a280-n279.txt")
TTP_instance = TTPSolver(TTP_inst)
n_pop = [5, 10, 20, 50, 100]
n_offspring = [10, 20, 50, 100, 200]
#data = EA_paramters_tuning(TTP_instance,n_pop=n_pop,n_offspring=n_offspring,problem = 'TTP',initial_strategy = 'random',local_search=True,verbose = True, repeat= 10)
plot(data, n_pop, n_offspring)
sub_data.append((solution.score, time() - start))
"""
data.append(sub_data)
# If you add a strategy, please add its label here:
strategies_labels = ("independent", "local_search", "annealing") #, "hybrid_EA/20/20")
data = np.array(data).transpose(1, 0, 2)
return strategies_labels, data
for instance_uri in instances_uris:
try:
instance = TTPSolver(TTPInstance(instance_uri))
except:
continue
name = instance_uri.split("/")[-1]
# Get a array of size (n_strategies x n_repeat x 2)
# The dimension of size 3 is for (length, time duration)
# The dimension of size n_repeat is for average and standard deviation purpose
strategies_labels, data = try_strategies(instance)
plt.boxplot(data[:, :, 0].T)
plt.xticks(range(1, len(strategies_labels) + 1), strategies_labels, rotation=45)
plt.title("score for: " + name)
plt.savefig("./experiment_data/overall_analysis/ttp/" + name + ".score.png", bbox_inches='tight')
plt.clf()