def plot_size_graph(response_size, size_stdev, action=None, tolerance=0):
plt.clf()
full_length = 100.0
step = 0.1
main_step = full_length / len(response_size)
x = np.arange(0, full_length, main_step)
# One main step less because levels go on sticks, not
# on intervals.
xmax = full_length - main_step + step
ymax = constants.n_labels
plt.errorbar(x,
response_size,
fmt='g-D',
yerr=size_stdev,
label=_('Average number of responses'))
plt.xlim(0, xmax)
# plt.ylim(0, ymax)
plt.xticks(x, constants.memory_sizes)
plt.yticks(np.arange(0, ymax + 1, 1), range(constants.n_labels + 1))
plt.xlabel(_('Range Quantization Levels'))
plt.ylabel(_('Size'))
plt.legend(loc=1)
plt.grid(True)
graph_filename = constants.picture_filename('graph_size_MEAN' +
_('-english'),
action,
tolerance=tolerance)
plt.savefig(graph_filename, dpi=600)
def plot_behs_graph(no_response,
no_correct,
no_chosen,
correct,
action=None,
tolerance=0):
for i in range(len(no_response)):
total = (no_response[i] + no_correct[i] + no_chosen[i] +
correct[i]) / 100.0
no_response[i] /= total
no_correct[i] /= total
no_chosen[i] /= total
correct[i] /= total
plt.clf()
full_length = 100.0
step = 0.1
main_step = full_length / len(constants.memory_sizes)
x = np.arange(0.0, full_length, main_step)
# One main step less because levels go on sticks, not
# on intervals.
xmax = full_length - main_step + step
ymax = full_length
width = 5 # the width of the bars: can also be len(x) sequence
plt.bar(x, correct, width, label=_('Correct response chosen'))
cumm = np.array(correct)
plt.bar(x,
no_chosen,
width,
bottom=cumm,
label=_('Correct response not chosen'))
cumm += np.array(no_chosen)
plt.bar(x, no_correct, width, bottom=cumm, label=_('No correct response'))
cumm += np.array(no_correct)
plt.bar(x, no_response, width, bottom=cumm, label=_('No responses'))
plt.xlim(-width, full_length + width)
plt.ylim(0.0, full_length)
plt.xticks(x, constants.memory_sizes)
plt.xlabel(_('Range Quantization Levels'))
plt.ylabel(_('Labels'))
plt.legend(loc=0)
plt.grid(axis='y')
graph_filename = constants.picture_filename('graph_behaviours_MEAN' +
_('-english'),
action,
tolerance=tolerance)
plt.savefig(graph_filename, dpi=600)
def plot_pre_graph (pre_mean, rec_mean, ent_mean, pre_std, rec_std, ent_std, \
tag = '', xlabels = constants.memory_sizes, xtitle = None, \
ytitle = None, action=None, occlusion = None, bars_type = None, tolerance = 0):
plt.clf()
plt.figure(figsize=(6.4, 4.8))
full_length = 100.0
step = 0.1
main_step = full_length / len(xlabels)
x = np.arange(0, full_length, main_step)
# One main step less because levels go on sticks, not
# on intervals.
xmax = full_length - main_step + step
# Gives space to fully show markers in the top.
ymax = full_length + 2
plt.errorbar(x, pre_mean, fmt='r-o', yerr=pre_std, label=_('Precision'))
plt.errorbar(x, rec_mean, fmt='b--s', yerr=rec_std, label=_('Recall'))
plt.xlim(0, xmax)
plt.ylim(0, ymax)
plt.xticks(x, xlabels)
if xtitle is None:
xtitle = _('Range Quantization Levels')
if ytitle is None:
ytitle = _('Percentage')
plt.xlabel(xtitle)
plt.ylabel(ytitle)
plt.legend(loc=4)
plt.grid(True)
entropy_labels = [str(e) for e in np.around(ent_mean, decimals=1)]
cmap = mpl.colors.LinearSegmentedColormap.from_list(
'mycolors', ['cyan', 'purple'])
Z = [[0, 0], [0, 0]]
levels = np.arange(0.0, xmax, step)
CS3 = plt.contourf(Z, levels, cmap=cmap)
cbar = plt.colorbar(CS3, orientation='horizontal')
cbar.set_ticks(x)
cbar.ax.set_xticklabels(entropy_labels)
cbar.set_label(_('Entropy'))
s = tag + 'graph_prse_MEAN' + _('-english')
graph_filename = constants.picture_filename(s, action, occlusion,
bars_type, tolerance)
plt.savefig(graph_filename, dpi=600)
def plot_features_graph(domain,
means,
stdevs,
experiment,
occlusion=None,
bars_type=None):
""" Draws the characterist shape of features per label.
The graph is a dots and lines graph with error bars denoting standard deviations.
"""
ymin = np.PINF
ymax = np.NINF
for i in constants.all_labels:
yn = (means[i] - stdevs[i]).min()
yx = (means[i] + stdevs[i]).max()
ymin = ymin if ymin < yn else yn
ymax = ymax if ymax > yx else yx
main_step = 100.0 / domain
xrange = np.arange(0, 100, main_step)
fmts = get_formats(constants.n_labels)
for i in constants.all_labels:
plt.clf()
plt.figure(figsize=(12, 5))
plt.errorbar(xrange,
means[i],
fmt=fmts[i],
yerr=stdevs[i],
label=str(i))
plt.xlim(0, 100)
plt.ylim(ymin, ymax)
plt.xticks(xrange, labels='')
plt.xlabel(_('Features'))
plt.ylabel(_('Values'))
plt.legend(loc='right')
plt.grid(True)
filename = constants.features_name(
experiment, occlusion, bars_type) + '-' + str(i) + _('-english')
plt.savefig(constants.picture_filename(filename), dpi=500)