def test_plot_bars_align_left(self):
actual = visualize.plot_bars(x=[1, 2], heights=[1, 2], label="yo", hatch="/")
expected = [
plt.Rectangle(
# if align == "center":
# rect_xmin = xpt - half_width
# rect_xmax = xpt + half_width
xy=(
1,
0,
),
height=1,
width=0.8, # width = diff * 0.8
alpha=0.8,
label="yo",
hatch="/",
),
plt.Rectangle(
# if align == "center":
# rect_xmin = xpt - half_width
# rect_xmax = xpt + half_width
xy=(
2,
0,
),
height=2,
width=0.8, # width = diff * 0.8
alpha=0.8,
hatch="/",
),
]
self._assert_matplotlib_rectangles_list(actual, expected)
def test_plot_bars_align_right(self):
actual = visualize.plot_bars(
x=[1, 2], heights=[1, 2], width=1, label="yo", hatch="/"
)
expected = [
plt.Rectangle(
xy=(
2,
0,
),
height=1,
width=1,
alpha=0.8,
label="yo",
hatch="/",
),
plt.Rectangle(
xy=(
3,
0,
),
height=2,
width=1,
alpha=0.8,
hatch="/",
),
]
self._assert_matplotlib_rectangles_list(actual, expected)
def psquare(self, x, y, col, full=False):
if full:
plt.gcf().gca().add_artist(
plt.Rectangle((x, y), self.Size, self.Size, ec=col, color=col))
else:
plt.gcf().gca().add_artist(
plt.Rectangle((x, y),
self.Size,
self.Size,
ec=col,
color='white'))
def line_select_callback(eclick, erelease):
x1, y1 = eclick.xdata, eclick.ydata
x2, y2 = erelease.xdata, erelease.ydata
rect = plt.Rectangle((min(x1, x2), min(y1, y2)), np.abs(x1 - x2),
np.abs(y1 - y2))
axes.add_patch(rect)
print('efd')
def plot_aperture_images(self, stack=None, info=None, lines=None):
if stack is None: stack = self.ap_stack
WQB = WQStackBrowser(self.ap_stack, info, lines, aspect='auto')
c = self.crop
WQB.axis.add_patch(plt.Rectangle((c['xmin'],c['ymin']), \
c['xmax']-c['xmin'],c['ymax']-c['ymin'],lw=3,ec='red',fc='0.5',alpha=0.2))
return WQB
def draw(self, ax, center, bottom, highlight_words={}):
if self.height is None or self.width is None or self.child_bottoms is None:
self.compute_size()
if self.is_leaf:
if self.contents in highlight_words:
self.text_properties['color'] = highlight_words[self.contents]
else:
self.text_properties['color'] = 'k'
ax.text(center,
bottom,
self.contents,
transform=None,
**self.text_properties)
else:
[
child.draw(ax, center, bottom + child_bottom,
highlight_words) for child, child_bottom in zip(
self.k_children, self.child_bottoms)
]
ax.add_patch(
plt.Rectangle((center - .5 * self.width, bottom),
self.width,
self.height - self.pts_buffer / 2,
alpha=.1,
transform=None))
ax.set_axis_off()
def wordshift_legend(ax, xoffset, yoffset, bwidth=1.05):
lineheight = 0.09
import matplotlib.transforms as transforms
trans = transforms.blended_transform_factory(ax.transAxes, ax.transAxes)
ax2 = plt.gcf().add_axes([0, 0, 1, 1])
ax2.axis('off')
ax2.xaxis.set_visible(False)
ax2.yaxis.set_visible(False)
ax2.set_zorder(1000)
patch = plt.Rectangle((xoffset - 0.025, yoffset - 4 * lineheight),
bwidth,
4.5 * lineheight,
facecolor=[0.9, 0.9, 0.9],
edgecolor='k',
linewidth=LINEWIDTH,
transform=trans)
ax2.add_patch(patch)
txtprops = {'va': 'top', 'transform': trans, 'fontsize': 'small'}
plt.text(xoffset, yoffset, u'+ High valence', color='orange', **txtprops)
plt.text(xoffset,
yoffset - 1 * lineheight,
u' - Low valence',
color='blue',
**txtprops)
plt.text(xoffset, yoffset - 2 * lineheight, u'$↑$ Overexpressed',
**txtprops)
plt.text(xoffset, yoffset - 3 * lineheight, u'$↓$ Underexpressed',
**txtprops)
def plot(self,circle=True,square=False,savefig=False,spath='../figs/eROSITA/'):
"""
plot cluster image with radius and mass information
"""
w_pix = self.w_pix
mid_pix = self.mid_pix
Rpixel = self.Rpixel
cluster = self.image
cluster_id = self.id
plt.figure(figsize=(4,4))
cmap = plt.cm.rainbow
cmap.set_under(color="white",alpha=0)
agn_cmap = plt.cm.viridis
agn_cmap.set_under(color="white",alpha=0)
im = plt.imshow(np.log10(self.image),cmap=cmap,interpolation='none')
#im = plt.imshow(np.log10(agn),cmap=agn_cmap,interpolation='none')
ax = plt.gca()
if self.meta is None:
plt.title('Noise')
else:
cluster_row = self.meta
log_m = np.log10(cluster_row['M500_msolh'])[0]
plt.title(r'$\log\left(\frac{M_\mathrm{500c}}{h^{-1}\,M_\odot}\right) = $'+'{:.2f}'.format(log_m))
if circle:
circle = plt.Circle((self.xmid, self.ymid), self.Rpixel, color="red",fill=False,zorder=2000)
ax.add_patch(circle)
if square:
circle = plt.Rectangle((self.xmid-384/2, self.ymid-384/2), 384, 384, color="black",fill=False,zorder=2000)
ax.add_patch(circle)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.12)
plt.colorbar(im, cax=cax)
ax.set_xticks(np.arange(0,w_pix,100))
ax.set_yticks(np.arange(0,w_pix,100))
ax.set_xlabel("x"), ax.set_ylabel("y")
ax.set_facecolor('white')
plt.tight_layout()
if savefig:
plt.savefig('fpath' + file_name + '.png',dpi=250,bbox_inches='tight')
ax.invert_yaxis()
plt.show()
plt.close()
return None
def draw_detection(img_path, label_path, savedir, num2label_map=None, label2num_map=None, absCoord=False):
img = imread(img_path)
h, w, c = img.shape
img_name = get_image_name(img_path)
bboxes = load_bboxes(label_path, w, h, num2label_map,absCoord)
plt.clf()
plt.imshow(img)
plt.axis('off');
ax = plt.gca()
num_classes = len(num2label_map) if num2label_map else 15
colors = plt.cm.hsv(np.linspace(0, 1, num_classes)).tolist()
for obj in bboxes:
label = obj['label']
color_num = label2num_map[label] if label2num_map else int(label)
color = colors[color_num % num_classes]
bbox = obj['bbox']
coords = (bbox[0] - bbox[2]/2., bbox[1]-bbox[3]/2.), bbox[2], bbox[3]
ax.add_patch(plt.Rectangle(*coords, fill=False, edgecolor=color, linewidth=1))
if 'score' in obj:
score = obj['score']
display_text = '%s: %.2f' % (label, score)
else:
display_text = str(label)
ax.text(coords[0][0], coords[0][1], display_text, bbox={'facecolor':color, 'alpha':0.5})
if savedir and bboxes:
save_img = os.path.join(savedir, img_name)
plt.savefig(save_img, bbox_inches="tight")
print "saving image:%s" % save_img
def plot_random_returns(df_price_oos, df_hard_returns, n=1000):
returns_length = df_price_oos.shape[0]
rand = np.random.random((n, returns_length))
rand_yn = rand > .5
rand_returns = rand_yn * df_price_oos.next_day_log_return.as_matrix()
returns = np.exp(np.cumsum(rand_returns, axis=1))
x = np.arange(rand_returns.shape[1])
with sns.axes_style("whitegrid"):
fig, ax = plt.subplots()
np.exp(df_price_oos.log_return.cumsum()).plot(ax=ax, color="black")
xlim = ax.get_xlim()
x = np.linspace(xlim[0], xlim[1], rand_returns.shape[1])
ps = []
labels = []
#p1 =
#p2 = Rectangle((0, 0), 1, 1, fc="red")
#legend([p1, p2], [a1_label, a2_label])
for alpha, q in [(.10, 1), (.20, 5), (.30, 10), (.40, 20), (.50, 35)]:
upper = np.percentile(returns, 100 - q, axis=0)
lower = np.percentile(returns, q, axis=0)
ps.append(plt.Rectangle((0, 0), 1, 1, fc="blue", alpha=alpha))
labels.append("{} - {}".format(
float(q) / 100, (100 - float(q)) / 100))
ax.fill_between(x, upper, lower, alpha=alpha)
np.exp(df_hard_returns[df_hard_returns.tail(1).idxmax(axis=1)]).plot(
ax=ax, color="red")
l1 = ax.legend(ps, labels, loc=2)
l2 = ax.legend(
["Bitcoin", df_hard_returns.tail(1).idxmax(axis=1)[0]], loc=3)
ax.add_artist(l1)
fig.savefig("thesis/plots/strat_random_returns.png")
def setPoints(fname, aviProps, bg):
frameWidth = aviProps[2]
frameHeight = aviProps[3]
fig, ax = showArena(aviProps, bg)
# 1 point for nest position
print "Set nest position [1 point]"
nestPosition = fig.ginput(1)
nestPosition = nestPosition[0]
ax.plot(nestPosition[0],nestPosition[1], '+')
fig.canvas.draw()
# 4 points for nest area
print "Set nest area [3 points: lower left, lower right, top]"
nestArea = fig.ginput(3)
W = nestArea[1][0] - nestArea[0][0]
H = nestArea[2][1] - nestArea[1][1]
rect = plt.Rectangle((nestArea[0][0],nestArea[0][1]),W,H, fc='r', alpha=0.2)
ax.add_patch(rect)
fig.canvas.draw()
# 1 point for middle of arena
print "Set centre of arena [1 point]"
arenaCentre = fig.ginput(1)
arenaCentre = arenaCentre[0]
x = [arenaCentre[0], arenaCentre[0]]
y = [0,frameHeight-60]
ax.plot(x,y, 'b')
fig.canvas.draw()
# 2 points for feeding area
print "Set feeding area [3 points: lower left, lower right, top]"
feedingArea = fig.ginput(3)
#circle = plt.Circle((feedingArea[0][0],feedingArea[0][1]), radius=feedingArea[1][0]-feedingArea[0][0], fc='c')
#ax.add_patch(circle)
W = feedingArea[1][0] - feedingArea[0][0]
H = feedingArea[2][1] - feedingArea[1][1]
rect2 = plt.Rectangle((feedingArea[0][0],feedingArea[0][1]),W,H, fc='c', alpha=0.2)
ax.add_patch(rect2)
fig.canvas.draw()
pts = [nestPosition, nestArea, arenaCentre, feedingArea]
#fsaveName = fname.rstrip(".avi") + "_arena"
#np.save(fsaveName, ths)
#print fsaveName, "saved"
return pts
def plot_box(bb, color='lightgreen'):
plt.gca().add_patch(
plt.Rectangle((bb[1], bb[0]),
bb[3] - bb[1],
bb[2] - bb[0],
facecolor='none',
edgecolor=color,
linewidth=2.0))
def color_box(vid, bbox, ax, color="red"):
ax.add_patch(
plt.Rectangle((int(bbox[0]), int(bbox[2])),
int(bbox[1] - bbox[0]),
int(bbox[3] - bbox[2]),
fill=False,
edgecolor=color,
linewidth=3.5))
def plot_generated_toy_batch(X_real, generator_model, discriminator_model, noise_dim, gen_iter, noise_scale=0.5):
# Generate images
X_gen = sample_noise(noise_scale, 10000, noise_dim)
X_gen = generator_model.predict(X_gen)
# Get some toy data to plot KDE of real data
data = load_toy(pts_per_mixture=200)
x = data[:, 0]
y = data[:, 1]
xmin, xmax = -1.5, 1.5
ymin, ymax = -1.5, 1.5
# Peform the kernel density estimate
xx, yy = np.mgrid[xmin:xmax:100j, ymin:ymax:100j]
positions = np.vstack([xx.ravel(), yy.ravel()])
values = np.vstack([x, y])
kernel = stats.gaussian_kde(values)
f = np.reshape(kernel(positions).T, xx.shape)
# Plot the contour
fig = plt.figure(figsize=(10,10))
plt.suptitle("Generator iteration %s" % gen_iter, fontweight="bold", fontsize=22)
ax = fig.gca()
ax.contourf(xx, yy, f, cmap='Blues', vmin=np.percentile(f,80), vmax=np.max(f), levels=np.linspace(0.25, 0.85, 30))
# Also plot the contour of the discriminator
delta = 0.025
xmin, xmax = -1.5, 1.5
ymin, ymax = -1.5, 1.5
# Create mesh
XX, YY = np.meshgrid(np.arange(xmin, xmax, delta), np.arange(ymin, ymax, delta))
arr_pos = np.vstack((np.ravel(XX), np.ravel(YY))).T
# Get Z = predictions
ZZ = discriminator_model.predict(arr_pos)
ZZ = ZZ.reshape(XX.shape)
# Plot contour
ax.contour(XX, YY, ZZ, cmap="Blues", levels=np.linspace(0.25, 0.85, 10))
dy, dx = np.gradient(ZZ)
# Add streamlines
# plt.streamplot(XX, YY, dx, dy, linewidth=0.5, cmap="magma", density=1, arrowsize=1)
# Scatter generated data
plt.scatter(X_gen[:1000, 0], X_gen[:1000, 1], s=20, color="coral", marker="o")
l_gen = plt.Line2D((0,1),(0,0), color='coral', marker='o', linestyle='', markersize=20)
l_D = plt.Line2D((0,1),(0,0), color='steelblue', linewidth=3)
l_real = plt.Rectangle((0, 0), 1, 1, fc="steelblue")
# Create legend from custom artist/label lists
# bbox_to_anchor = (0.4, 1)
ax.legend([l_real, l_D, l_gen], ['Real data KDE', 'Discriminator contour',
'Generated data'], fontsize=18, loc="upper left")
ax.set_xlim(xmin, xmax)
ax.set_ylim(ymin, ymax + 0.8)
plt.savefig("../../figures/toy_dataset_iter%s.jpg" % gen_iter)
plt.clf()
plt.close()
def plot_screen(tiles, gazepoints, tolerance=0, filepath=None):
""" plot an episode screen
tiles -- a list of tile objects
gaze points -- a series of sfvec2f objects
filepath -- optional argument to save the plot to
a file. The filepath MUST be have a
valid image extension, e.g. "file.png"
(bmp, jpg, etc)
"""
#read gaze points
gazeX = gazepoints['pixels'].map(lambda x: x.x)
gazeY = gazepoints['pixels'].map(lambda x: x.y)
gazeI = gazepoints.index
#create figure
fig = plt.figure(figsize=(16, 9))
ax = fig.add_subplot(111)
ax.set_xlim(-200, 2120)
ax.set_ylim(-200, 1280)
ax.set_ylim(ax.get_ylim()[::-1])
ax.add_patch(plt.Rectangle((0, 0), 1920, 1080, alpha=0.05, color='b'))
gax = ax.scatter(gazeX,
gazeY,
s=10,
c=gazeI,
cmap=plt.cm.Reds,
label='gaze')
fig.colorbar(gax, format='%d')
#draw the objects
tolerances = []
for tile in tiles:
if tile.is_selected:
ax.text(tile.x,
tile.y,
tile.shape,
size='xx-large',
weight='bold',
color=tile.color,
bbox=dict(facecolor='red', alpha=0.5))
else:
ax.text(tile.x,
tile.y,
tile.shape,
size='xx-large',
weight='bold',
color=tile.color)
#draw the tolerance circles
circle = plt.Circle((tile.x, tile.y), tolerance, color='r', fill=False)
ax.add_patch(circle)
if filepath is not None:
fig.savefig(filepath, format=filepath.split('.')[1])
def mark_period(ax, xl, ec='k', Linewidth=1, fill=False, fc=(1, 1, 1, .5)):
""" plots a rectagular box in plot with height according to axis given """
yl = ax.get_ylim()
p = plt.Rectangle((xl[0], yl[0]),
np.diff(xl),
np.diff(yl),
edgecolor=ec,
fill=fill,
Linewidth=Linewidth,
facecolor=fc)
ax.add_patch(p)
def plotArena(aviProps, pmts, bg):
frameWidth = aviProps[2]
frameHeight = aviProps[3]
fig, ax = showArena(aviProps, bg)
fig.show()
nestPosition, nestArea, arenaCentre, feedingArea = pmts[2], pmts[3], pmts[4], pmts[5]
ax.plot(nestPosition[0],nestPosition[1], '+')
W = nestArea[1][0] - nestArea[0][0]
H = nestArea[2][1] - nestArea[1][1]
rect = plt.Rectangle((nestArea[0][0],nestArea[0][1]),W,H, fc='r', alpha=0.2)
ax.add_patch(rect)
x = [arenaCentre[0], arenaCentre[0]]
y = [0,frameHeight-60]
ax.plot(x,y, 'b')
W = feedingArea[1][0] - feedingArea[0][0]
H = feedingArea[2][1] - feedingArea[1][1]
rect2 = plt.Rectangle((feedingArea[0][0],feedingArea[0][1]),W,H, fc='c', alpha=0.2)
ax.add_patch(rect2)
fig.canvas.draw()
def slide_7_2():
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
rect = plt.Rectangle((0.2, 0.75), 0.4, 0.15, color='k', alpha=0.3)
circ = plt.Circle((0.7, 0.2), 0.15, color='b', alpha=0.3)
pgon = plt.Polygon([[0.15, 0.15], [0.35, 0.4], [0.2, 0.6]],
color='g',
alpha=0.5)
ax.add_patch(rect)
ax.add_patch(circ)
ax.add_patch(pgon)
def plot_image(fileobj, filename=None, show_corrects=False, bare=False):
img = gv.img.load_image(fileobj.path)
fig = plt.figure()
ax = fig.add_subplot(111)
s = dict(cmap=plt.cm.gray) if img.ndim == 2 else {}
ax.imshow(img, **s)
if bare:
ax.set_xticks([])
ax.set_yticks([])
for bbobj in fileobj.boxes:
bb = bbobj.box
if show_corrects:
if bbobj.correct:
color = 'lightgreen'
else:
color = 'red'
else:
if bbobj.difficult:
color = 'red'
else:
color = 'lightgreen'
plt.gca().add_patch(
plt.Rectangle((bb[1], bb[0]),
bb[3] - bb[1],
bb[2] - bb[0],
facecolor='none',
edgecolor=color,
linewidth=2.0))
if not bare:
plt.text(bb[1],
bb[0],
"{0:.2f}".format(bbobj.confidence),
color='white',
size=6,
ha='left',
va='bottom')
#plt.gca().add_patch(plt.Rectangle((bb[0], bb[1]), bb[2]-bb[0], bb[3]-bb[1], facecolor='none', edgecolor='lightgreen', linewidth=2.0))
if not bare:
ax.set_title("img_id = {0}".format(fileobj.img_id))
else:
plt.tight_layout()
if filename is not None:
plt.savefig(filename)
else:
plt.show()
def hinton(matrix, max_weight=None, ax=None):
"""Draw Hinton diagram for visualizing a weight matrix."""
ax = ax if ax is not None else plt.gca()
if not max_weight:
max_weight = 2 ** np.ceil(np.log(np.abs(matrix).max()) / np.log(2))
ax.patch.set_facecolor('gray')
ax.set_aspect('equal', 'box')
ax.xaxis.set_major_locator(plt.NullLocator())
ax.yaxis.set_major_locator(plt.NullLocator())
for (x, y), w in np.ndenumerate(matrix):
color = 'white' if w > 0 else 'black'
size = np.sqrt(np.abs(w) / max_weight)
rect = plt.Rectangle([x - size / 2, y - size / 2], size, size,
facecolor=color, edgecolor=color)
ax.add_patch(rect)
ax.autoscale_view()
ax.invert_yaxis()
def show_item(item):
img, bbox, label, scale = item
img = img.transpose(1, 2, 0)
img = denormalize(img)
title = 'Label: ' + str(label) + " shape: " + str(img.shape)
plt.figure()
plt.title(title)
plt.imshow(img)
ax = plt.gca()
for i in range(bbox.shape[0]):
xy = (bbox[i, 1], bbox[i, 0])
w, h = bbox[i, 3] - bbox[i, 1], bbox[i, 2] - bbox[i, 0]
rect = plt.Rectangle(xy,
w,
h,
fill=False,
edgecolor='red',
linewidth=1)
ax.add_patch(rect)
plt.axis('off')
plt.show()
def box(x=None, y=None, normal=False, fill=True, ax=None, **kwargs):
oline = kwargs.pop('outline', False)
outline_prop = kwargs.pop('outline_prop', {})
tmp = dict(
color='0.2',
alpha=0.4,
linewidth=2,
)
if ax is None:
ax = pylab.gca()
axt = ax.axis()
if normal:
tmp['transform'] = ax.transAxes
else:
tmp['transform'] = ax.transData
# if x is not None and percent: x = np.diff(axt[:2])*np.array(x) + axt[0]
# if y is not None and percent: y = np.diff(axt[2:])*np.array(y) + axt[2]
if x is None: x = axt[:2]
if y is None: y = axt[2:]
dx = np.diff(x)[0]
dy = np.diff(y)[0]
# use date2num
# if isinstance(x[0], datetime): dx = dx.days
# if isinstance(y[0], datetime): dy = dy.total_seconds()
if oline or (not fill):
tmp['facecolor'] = 'none'
tmp['edgecolor'] = kwargs.pop('color', tmp.pop('color'))
tmp.update(kwargs)
patch = pylab.Rectangle((x[0], y[0]), dx, dy, **tmp)
if oline:
outline(patch, **outline_prop)
ax.add_patch(patch)
del iris_data['petal_width']
# Create scatterplot of sepal_length by sepal_width colored by class
color_dict = {
'Iris-setosa': 'red',
'Iris-versicolor': 'blue',
'Iris-virginica': 'green'
}
plt.xlabel('Sepal Length')
plt.ylabel('Sepal Width')
plt.title('Iris Classification by Sepal')
plt.scatter(iris_data['sepal_length'],
iris_data['sepal_width'],
color=[color_dict[i] for i in iris_data['iris_class']])
plt.grid()
p1 = plt.Rectangle((0, 0), 0.1, 0.1, fc=color_dict['Iris-setosa'])
p2 = plt.Rectangle((0, 0), 0.1, 0.1, fc=color_dict['Iris-versicolor'])
p3 = plt.Rectangle((0, 0), 0.1, 0.1, fc=color_dict['Iris-virginica'])
plt.legend((p1, p2, p3), ('Iris-setosa', 'Iris-versicolor', 'Iris-virginica'),
loc='best')
#plt.show()
# Random point for plotting
random_x = random.uniform(4.0, 8.5)
random_y = random.uniform(1.5, 5.0)
# measures the distance from random point against all points in data frame
# records info in data frame column "distance"
for i in iris_data.index:
train_x = iris_data['sepal_length'][i]
train_y = iris_data['sepal_width'][i]
title,
horizontalalignment='center',
fontsize=35,
transform=ax.transAxes)
ax.tick_params(axis='x', labelsize=15)
plt.xticks(rotation=70)
xticks = ax.xaxis.get_major_ticks()
xticks[0].label1.set_visible(False)
ax.tick_params(axis='y', labelsize=15)
#ax.xaxis.set_label_coords(1.00, -0.03)
#ax.margins(0.1, 0)
plt.xlim(0, 1)
#y_max = Y1.max() * 1.2
plt.ylim((-0.01, 0.1))
p = []
for i in color:
p.extend([plt.Rectangle((0, 0), 1, 1, fc=i, alpha=1)])
lgd = ax.legend(p, words, loc='center', bbox_to_anchor=(1.15, 0.5))
path = '/Users/gerauddaspremont/Dropbox/project/thesis_1/Figures'
filename = 'lambdavsFreq2' + w.replace(' ', '') + '.pdf'
filename = os.path.join(path, filename)
fig.savefig(filename, bbox_inches='tight')
def plot_bars( # noqa
x, heights, width=None, align="center", y_base=0, ax=None, **rectangle_kws
): # pylint: disable=too-many-arguments,too-many-locals,too-many-branches,too-many-statements,line-too-long # noqa
"""Plot rectangular bars.
Args:
x (List): X points of the bars. Can handle dates.
heights (List): Heights of the bars, can be negative.
width (float or Any): Used with the center to determine the bar
dimensions.
align (str): Where to align the x.
'center': x is the center.
'left': x is the left edge.
'right': x is the right edge.
y_base (float): Base of the bars.
ax (matplotlib.axes.Axes): Axis object, otherwise uses plt.gca().
**rectangle_kws: passed to all plt.Rectangle(**rectangle_kws).
Returns:
List[plt.Rectangle]: List of the rectangle patches.
"""
ax = ax or plt.gca()
if len(x) == 0:
# don't plot anything because there isn't any
return []
# is x axis datetime objects?
xpt0 = x[0]
x_is_datetime = isinstance(xpt0, datetime)
# get width of the bars
if width is None:
if len(x) > 1:
xpt1 = x[1]
diff = xpt1 - xpt0
width = diff * 0.8
else:
if x_is_datetime:
width = timedelta(days=1)
else:
width = 1
elif isinstance(width, (float, int)) and x_is_datetime:
width = timedelta(days=width)
half_width = width / 2.0
# get the x and y bounds
xmin, xmax = ax.get_xlim()
ymin, ymax = ax.get_ylim()
is_first = True
rectangles = []
for xpt, height in zip(x, heights):
if height < 0:
height = abs(height)
rect_ymin = y_base - height
rect_ymax = y_base
else:
rect_ymin = y_base
rect_ymax = rect_ymin + height
if align == "center":
rect_xmin = xpt - half_width
rect_xmax = xpt + half_width
elif align == "left":
rect_xmin = xpt
rect_xmax = xpt + width
elif align == "right":
rect_xmin = xpt - width
rect_xmax = xpt
else:
raise ValueError(
f'unknown value for where "{align}" not one of (center, right, left)'
)
kws = {
"alpha": 0.8,
}
kws.update(rectangle_kws)
if not is_first:
kws.pop("label", None)
rect = plt.Rectangle(
xy=(rect_xmin, rect_ymin),
height=height,
width=width,
**kws,
)
ax.add_patch(rect)
rectangles.append(rect)
if isinstance(rect_xmin, datetime) and isinstance(xmin, float):
if xmin == 0.0:
xmin = rect_xmin
else:
xmin = plt.matplotlib.dates.num2date(xmin).replace(
tzinfo=rect_xmin.tzinfo
)
if isinstance(rect_xmax, datetime) and isinstance(xmax, float):
if xmax == 1.0:
xmax = rect_xmax
else:
xmax = plt.matplotlib.dates.num2date(xmax).replace(
tzinfo=rect_xmax.tzinfo
)
xmin = min(xmin, rect_xmin)
xmax = max(xmax, rect_xmax)
ymin = min(ymin, rect_ymin)
ymax = max(ymax, rect_ymax)
is_first = False
ax.set_xlim(xmin, xmax)
ax.set_ylim(ymin, ymax)
return rectangles
def plot_percentiles(self,
ax=None,
box_width=0.2,
wisker_size=20,
mean_size=10,
median_size=10,
line_width=1.5,
xoffset=0,
color=0,
tickbase=1):
"""
This will plot the percentiles of the data ([5, 25, 50, 75, 95]. In addition, the mean (o) and median (_) are shown.
Parameters
----------
ax
box_width
wisker_size
mean_size
median_size
line_width
xoffset
color
tickbase
Returns
-------
f, a, boxes, vlines, wisker_tips, mean
"""
if type(color) == int:
color = _plt.rcParams['axes.prop_cycle'].by_key()['color'][color]
col = _plt_tools.colors.Color(color, model='hex')
elif type(color) == str:
col = _plt_tools.colors.Color(color, model='hex')
else:
col = _plt_tools.colors.Color(color, model='rgb')
col.saturation = 0.3
color_bright = col.rgb
if ax:
a = ax
f = a.get_figure()
else:
f, a = _plt.subplots()
boxes = []
vlines = []
xordinal = []
for row in self.percentiles.iterrows():
# width = 10
x = row[0] + xoffset
xordinal.append(x)
# box
# y = (row[1][75] + row[1][25]) / 2
y = row[1][25]
height = row[1][75] - row[1][25]
box = _plt.Rectangle(
(x - box_width / 2, y),
box_width,
height,
# ha = 'center'
)
box.set_facecolor([1, 1, 1, 1])
a.add_patch(box)
boxes.append(box)
# wiskers
y = (row[1][95] + row[1][5]) / 2
vl = a.vlines(x, row[1][5], row[1][95])
vlines.append(vl)
for b in boxes:
b.set_linewidth(line_width)
b.set_facecolor(color_bright)
b.set_edgecolor(color)
b.set_zorder(2)
for vl in vlines:
vl.set_color(color)
vl.set_linewidth(line_width)
vl.set_zorder(1)
# wm_lw = 2
wisker_tips = []
if wisker_size:
g, = a.plot(xordinal, self.percentiles[5], ls='')
wisker_tips.append(g)
g, = a.plot(xordinal, self.percentiles[95], ls='')
wisker_tips.append(g)
for wt in wisker_tips:
wt.set_markeredgewidth(line_width)
wt.set_color(color)
wt.set_markersize(wisker_size)
wt.set_marker('_')
mean = None
if mean_size:
g, = a.plot(xordinal, self.percentiles['mean'], ls='')
g.set_marker('o')
g.set_markersize(mean_size)
g.set_zorder(20)
g.set_markerfacecolor('None')
g.set_markeredgewidth(line_width)
g.set_markeredgecolor(color)
mean = g
median = None
if median_size:
g, = a.plot(xordinal, self.percentiles['median'], ls='')
g.set_marker('_')
g.set_markersize(median_size)
g.set_zorder(20)
g.set_markeredgewidth(line_width)
g.set_markeredgecolor(color)
median = g
# a.xaxis.set_major_locator(_MonthLocator())
# a.xaxis.set_major_formatter(_DateFormatter('%b'))
try:
a.set_ylim(_np.nanmin(self.percentiles.drop(['n_valid'], axis=1)),
_np.nanmax(self.percentiles.drop(['n_valid'], axis=1)))
except:
pass
try:
a.set_xlim(self.percentiles.index.min(),
self.percentiles.index.max())
except:
pass
mjl = _MultipleLocator(tickbase)
a.xaxis.set_major_locator(mjl)
# a.relim()
# a.autoscale_view(tight=True)
# f.autofmt_xdate()
return f, a, boxes, vlines, wisker_tips, mean, median
def get_border_points(img, rel_threshold=0.1, dev_bg=2, xmin=0, xmax=np.inf, ymin=0, ymax=np.inf, \
medfilt_radius=11, window_size=None, interp_out=False, verbosity=1, **kwargs):
"""
Estimate x coordinates of the aperture border in wq-maps
for each y-line of the (previously binned) image.
img ... 2D array image (Ny,Nx), binned along y
rel_threshold ... (opt) relative threshold
dev ... (opt) deviation from linear fit w.r.t. std.dev.
dev_bg ... (opt) background noise level w.r.t. sdt.dev of noise
medfilt_radius... (opt) filtering of input signal
xmin,xmax ... (opt) restricts image for fitting (px)
ymin,ymax
window_size ... (opt) restrict to window around max intensity (px)
interp_out ... (opt) bool indicating, if outliers (NaN's) shall be
replaced with interpolated values (
verbosity ... (opt) verbosity (0=silent, 2=plotting fit, 3=debug)
RETURNS left and right border as list of points (x,y) with length ymax-ymin
"""
# image axis order (y,x)
xmax = min(xmax, img.shape[1])
ymax = min(ymax, img.shape[0])
xmin = max(xmin, 0)
ymin = max(ymin, 0)
Nx = xmax - xmin
Ny = ymax - ymin
assert Nx > 0 and Ny > 0
xleft = np.empty(Ny)
xright = np.empty(Ny)
x = np.arange(xmin, xmax)
# 1. try to find left and right border for each line y
# in the cropped image, i.e., ymin <= y < ymax, xmin <= x < xmax
for iy, y in enumerate(np.arange(ymin, ymax)):
# smoothing and background subtraction
line = medfilt1D(img[y, xmin:xmax], medfilt_radius)
out = outliers(x, line, iterations=3, dev=2)
bg = line[~out]
# linear background signal
line -= np.mean(bg)
# window around max intensity
ixcenter = np.argmax(line)
if window_size is not None:
ixmin = max(ixcenter - window_size / 2, 0)
ixmax = min(ixcenter + window_size / 2, Nx - 1)
else:
ixmin = 0
ixmax = Nx - 1
# find first point from left/right above threshold
thresh = line[ixcenter] * rel_threshold
if thresh < dev_bg * np.std(
bg): # S/N too low (threshold below std.dev)
xleft[iy] = xright[iy] = np.nan
else:
# image coordinates !
xleft[iy] = x[ixmin + threshold(line[ixmin:ixcenter], thresh)]
xright[iy] = x[ixmax - threshold(line[ixmax:ixcenter:-1], thresh)]
# DEBUG
if verbosity > 10 and iy == (ymax + ymin) / 2:
plt.figure()
plt.title("line at y = %d" % (Ny / 2))
plt.plot(img[y], 'k-', label="intensity")
plt.plot(x[~out], bg, 'b-', label="background")
plt.axhline(thresh + np.mean(bg), color='g')
plt.axvline(xleft[iy], color='r')
plt.axvline(xright[iy], color='r', label="border")
plt.legend(loc=2)
# 2. determine outliers
y = np.arange(ymin, ymax)
iout_left = outliers(y, xleft, **kwargs)
iout_right = outliers(y, xright, **kwargs)
# 3. plotting
if verbosity > 2:
plt.figure()
plt.title("Debug: Edge fitting")
plt.xlabel("x [px]")
plt.ylabel("y [px]")
plt.imshow(np.log(1 + np.abs(img)), cmap=plt.cm.gray, aspect='auto')
#plt.imshow(img,vmin=0,vmax=np.mean(img)*10,cmap=plt.cm.gray, aspect='auto');
plt.plot(xleft, y, 'g+', mew=2, label='border')
plt.plot(xright, y, 'g+', mew=2)
plt.plot(xleft[iout_left], y[iout_left], 'r+', mew=2, label='outliers')
plt.plot(xright[iout_right], y[iout_right], 'r+', mew=2)
plt.gca().add_patch(
plt.Rectangle((xmin, ymin),
Nx,
Ny,
lw=3,
ec='red',
fc='0.5',
alpha=0.2))
plt.xlim(0, img.shape[1])
plt.ylim(img.shape[0] - 1, 0)
plt.legend(loc=2)
# 4. calculate fitted points at outliers
xleft[iout_left] = np.nan
xright[iout_right] = np.nan
if interp_out:
xleft = replace_nan(y, xleft)
xright = replace_nan(y, xright)
# 5. return border as list of (x,y)-points
return np.vstack((xleft, y)), np.vstack((xright, y))
def draw_region(start, end, height, **kwargs):
""" Draw a rectangular patch that spans coordinates start to end, with given height.
"""
r = pylab.Rectangle((start, 0), abs(start-end), height, **kwargs)
pylab.gca().add_patch(r)
pylab.draw()
def cross_correl_plot(data, FIGSIZE=(7,7), wspace=.5, hspace=.5, right=0.98, left=0.1,\
many_data=False):
"""
'data' should be an array of dictionaries with keys 'vec' and labels 'label'
"""
fig, AX = plt.subplots(len(data) - 1, len(data) - 1, figsize=FIGSIZE)
plt.subplots_adjust(wspace=wspace, hspace=hspace, right=right, left=left)
mymap = get_linear_colormap(color1='white', color2='gray')
significance = np.array([1e-3, 2e-2, 1e-1, 1])
for i in range(len(data) - 1):
for j in range(i + 1, len(data)):
AX[j - 1, i].plot(data[i]['vec'], data[j]['vec'], 'ko')
if not many_data:
set_plot(AX[j - 1, i],
xlabel=data[i]['label'],
ylabel=data[j]['label'])
else:
if ((i == 0) and (j == len(data) - 1)):
set_plot(AX[j - 1, i],
xlabel=data[i]['label'],
ylabel=data[j]['label'])
elif (j == len(data) - 1):
set_plot(AX[j - 1, i],
xlabel=data[i]['label'],
yticks_labels=[])
elif (i == 0):
set_plot(AX[j - 1, i],
ylabel=data[j]['label'],
xticks_labels=[])
else:
set_plot(AX[j - 1, i], xticks_labels=[], yticks_labels=[])
cc, pp = pearsonr(data[i]['vec'], data[j]['vec'])
x = np.linspace(data[i]['vec'].min(), data[i]['vec'].max())
AX[j-1, i].plot(x,\
np.polyval(np.polyfit(np.array(data[i]['vec'], dtype='f8'),\
np.array(data[j]['vec'], dtype='f8'), 1), x),\
'k--', lw=.5)
ii = np.arange(len(significance))[significance - pp >= 0][0]
color = -1. * (ii - len(significance) + 1) / (len(significance) -
1)
xmin, xmax = AX[j - 1, i].get_xaxis().get_view_interval()
ymin, ymax = AX[j - 1, i].get_yaxis().get_view_interval()
AX[j-1, i].add_patch(plt.Rectangle((xmin, ymin),\
xmax-xmin, ymax-ymin, color=mymap(1.*color,1)))
ax = plt.axes([.7, .7, .02, .2])
build_bar_legend(np.arange(len(significance)+1),\
ax, mymap,\
ticks_labels=['n.s.', '$<$0.1', '$<$0.02', '$<$0.001'],
label='Significance \n \n (p, Pearson correl.)')
for ax in AX.flatten():
if ax.get_xaxis().get_view_interval()[1] == 1.:
ax.axis('off')
return fig
def plot_policy(env, args, x_range, y_range, policy, prefix):
"""
Plots 2d graph depicting a policy. Right now I don't have any mechanism to break ties, I expect the user to have
done that already. TODO: I don't support two actions at the same time, so I'll always depict only one action. I
should support this in the future.
:param env: object containing the environment's features, I access it to plot what I want
:param args: list of all arguments passed at the execution so we can access the paths we are interested at
:param x_range: one of the dimensions of the grid world
:param y_range: the other dimension of the grid world
:param policy: actual policy to be plotted. It should be encoded as an integer per state.
:param prefix: string that will be used as the filename of the generated graph
"""
plt.clf()
plt.close()
from pylab import rcParams
rcParams['figure.figsize'] = y_range, x_range
plt.xlim([0, y_range])
plt.ylim([0, x_range])
for i in range(y_range):
plt.axvline(i, color='k', linestyle=':')
plt.axvline(y_range, color='k', linestyle=':')
for j in range(x_range):
plt.axhline(j, color='k', linestyle=':')
plt.axhline(x_range, color='k', linestyle=':')
for idx in range(len(policy)):
i, j = env.get_state_xy(idx)
dx = 0
dy = 0
if policy[idx] == 0: # up
dy = 0.001
elif policy[idx] == 1: # right
dx = 0.001
elif policy[idx] == 2: # down
dy = -0.001
elif policy[idx] == 3: # left
dx = -0.001
if (env._matrix_mdp[i][j] != -1
and policy[idx] == 4) or (env._goal_x == i
and env._goal_y == j): # termination
termination = plt.Rectangle((j, x_range - i - 1), 1, 1, color='r')
plt.gca().add_artist(termination)
elif env._matrix_mdp[i][j] != -1:
plt.arrow(j + 0.5 - 250 * dx,
x_range - i + 0.5 - 1 - 250 * dy,
dx,
dy,
head_width=0.2,
head_length=0.5,
fc='k',
ec='k')
else:
plt.gca().add_patch(
patches.Rectangle(
(j, x_range - i - 1), # (x,y)
1.0, # width
1.0, # height
facecolor="gray"))
plt.savefig(args.output + prefix + '.png')
plt.clf()