def plot_optCurve_nEvals(self, plot_best=True, scale='log', type='mean+65'):
"""
assume that the experiment have the same lenght
:return:
"""
data = []
for idx, experiment in enumerate(self.experiments):
max_lenght = 0
for rep in np.arange(experiment.repetitions):
max_lenght = max(max_lenght, self._logs.logs[idx, rep].get_n_evals())
temp = np.nan * np.empty((experiment.repetitions, max_lenght)) # initialize matrix
for rep in np.arange(experiment.repetitions):
if plot_best:
temp[rep, 0:self._logs.logs[idx, rep].get_n_evals()] = bestof(np.matrix(self._logs.logs[idx, rep].get_objectives()).T).T
else:
temp[rep, 0:self._logs.logs[idx, rep].get_n_evals()] = self._logs.logs[idx, rep].get_objectives()
data.append(temp)
#
plt.figure()
spp.rplot_data(data, legend=self.get_labels(), typeplot=type)
plt.xlabel('Evaluations')
plt.ylabel('Obj.Func.')
if scale == 'log':
try:
ax = plt.gca()
ax.set_yscale('log')
except:
print('log scale is not possible')
def test_rplot_data_yscale(self):
y = [np.absolute(np.random.rand(100, 100)) + 0.00001]
x = np.random.rand(100)
h = spp.rplot_data(data=y)
plt.yscale('log')
def test_rplot_data_matrix(self):
y = np.random.rand(100, 100)
h = spp.rplot_data(data=y)
def test_rplot_data_3(self):
y = [np.random.rand(100, 100)]
x = np.random.rand(100)
h = spp.rplot_data(data=y)