def scatter(self,
x,
y,
xerr=None,
yerr=None,
cov=None,
corr=None,
s_expr=None,
c_expr=None,
labels=None,
selection=None,
length_limit=50000,
length_check=True,
label=None,
xlabel=None,
ylabel=None,
errorbar_kwargs={},
ellipse_kwargs={},
**kwargs):
"""Viz (small amounts) of data in 2d using a scatter plot
Convenience wrapper around pylab.scatter when for working with small DataFrames or selections
:param x: Expression for x axis
:param y: Idem for y
:param s_expr: When given, use if for the s (size) argument of pylab.scatter
:param c_expr: When given, use if for the c (color) argument of pylab.scatter
:param labels: Annotate the points with these text values
:param selection: Single selection expression, or None
:param length_limit: maximum number of rows it will plot
:param length_check: should we do the maximum row check or not?
:param label: label for the legend
:param xlabel: label for x axis, if None .label(x) is used
:param ylabel: label for y axis, if None .label(y) is used
:param errorbar_kwargs: extra dict with arguments passed to plt.errorbar
:param kwargs: extra arguments passed to pylab.scatter
:return:
"""
import pylab as plt
x = _ensure_strings_from_expressions(x)
y = _ensure_strings_from_expressions(y)
label = str(label or selection)
selection = _ensure_strings_from_expressions(selection)
if length_check:
count = self.count(selection=selection)
if count > length_limit:
raise ValueError(
"the number of rows (%d) is above the limit (%d), pass length_check=False, or increase length_limit"
% (count, length_limit))
x_values = self.evaluate(x, selection=selection)
y_values = self.evaluate(y, selection=selection)
if s_expr:
kwargs["s"] = self.evaluate(s_expr, selection=selection)
if c_expr:
kwargs["c"] = self.evaluate(c_expr, selection=selection)
plt.xlabel(xlabel or self.label(x))
plt.ylabel(ylabel or self.label(y))
s = plt.scatter(x_values, y_values, label=label, **kwargs)
if labels:
label_values = self.evaluate(labels, selection=selection)
for i, label_value in enumerate(label_values):
plt.annotate(label_value, (x_values[i], y_values[i]))
xerr_values = None
yerr_values = None
if cov is not None or corr is not None:
from matplotlib.patches import Ellipse
sx = self.evaluate(xerr, selection=selection)
sy = self.evaluate(yerr, selection=selection)
if corr is not None:
sxy = self.evaluate(corr, selection=selection) * sx * sy
elif cov is not None:
sxy = self.evaluate(cov, selection=selection)
cov_matrix = np.zeros((len(sx), 2, 2))
cov_matrix[:, 0, 0] = sx**2
cov_matrix[:, 1, 1] = sy**2
cov_matrix[:, 0, 1] = cov_matrix[:, 1, 0] = sxy
ax = plt.gca()
ellipse_kwargs = dict(ellipse_kwargs)
ellipse_kwargs['facecolor'] = ellipse_kwargs.get('facecolor', 'none')
ellipse_kwargs['edgecolor'] = ellipse_kwargs.get('edgecolor', 'black')
for i in range(len(sx)):
eigen_values, eigen_vectors = np.linalg.eig(cov_matrix[i])
indices = np.argsort(eigen_values)[::-1]
eigen_values = eigen_values[indices]
eigen_vectors = eigen_vectors[:, indices]
v1 = eigen_vectors[:, 0]
v2 = eigen_vectors[:, 1]
varx = cov_matrix[i, 0, 0]
vary = cov_matrix[i, 1, 1]
angle = np.arctan2(v1[1], v1[0])
# round off errors cause negative values?
if eigen_values[1] < 0 and abs(
(eigen_values[1] / eigen_values[0])) < 1e-10:
eigen_values[1] = 0
if eigen_values[0] < 0 or eigen_values[1] < 0:
raise ValueError('neg val')
width, height = np.sqrt(np.max(eigen_values)), np.sqrt(
np.min(eigen_values))
e = Ellipse(xy=(x_values[i], y_values[i]),
width=width,
height=height,
angle=np.degrees(angle),
**ellipse_kwargs)
ax.add_artist(e)
else:
if xerr is not None:
if _issequence(xerr):
assert len(
xerr
) == 2, "if xerr is a sequence it should be of length 2"
xerr_values = [
self.evaluate(xerr[0], selection=selection),
self.evaluate(xerr[1], selection=selection)
]
else:
xerr_values = self.evaluate(xerr, selection=selection)
if yerr is not None:
if _issequence(yerr):
assert len(
yerr
) == 2, "if yerr is a sequence it should be of length 2"
yerr_values = [
self.evaluate(yerr[0], selection=selection),
self.evaluate(yerr[1], selection=selection)
]
else:
yerr_values = self.evaluate(yerr, selection=selection)
if xerr_values is not None or yerr_values is not None:
errorbar_kwargs = dict(errorbar_kwargs)
errorbar_kwargs['fmt'] = errorbar_kwargs.get('fmt', 'none')
plt.errorbar(x_values,
y_values,
yerr=yerr_values,
xerr=xerr_values,
**errorbar_kwargs)
return s
def label(index, label, expression):
if label and _issequence(label):
return label[i]
else:
return self.label(expression)
def label(index, label, expression):
if label and _issequence(label):
return label[i]
else:
return self.label(expression)
def scatter(self, x, y, xerr=None, yerr=None, cov=None, corr=None, s_expr=None, c_expr=None, labels=None, selection=None, length_limit=50000,
length_check=True, label=None, xlabel=None, ylabel=None, errorbar_kwargs={}, ellipse_kwargs={}, **kwargs):
"""Viz (small amounts) of data in 2d using a scatter plot
Convenience wrapper around pylab.scatter when for working with small DataFrames or selections
:param x: Expression for x axis
:param y: Idem for y
:param s_expr: When given, use if for the s (size) argument of pylab.scatter
:param c_expr: When given, use if for the c (color) argument of pylab.scatter
:param labels: Annotate the points with these text values
:param selection: Single selection expression, or None
:param length_limit: maximum number of rows it will plot
:param length_check: should we do the maximum row check or not?
:param label: label for the legend
:param xlabel: label for x axis, if None .label(x) is used
:param ylabel: label for y axis, if None .label(y) is used
:param errorbar_kwargs: extra dict with arguments passed to plt.errorbar
:param kwargs: extra arguments passed to pylab.scatter
:return:
"""
import pylab as plt
x = _ensure_strings_from_expressions(x)
y = _ensure_strings_from_expressions(y)
label = str(label or selection)
selection = _ensure_strings_from_expressions(selection)
if length_check:
count = self.count(selection=selection)
if count > length_limit:
raise ValueError("the number of rows (%d) is above the limit (%d), pass length_check=False, or increase length_limit" % (count, length_limit))
x_values = self.evaluate(x, selection=selection)
y_values = self.evaluate(y, selection=selection)
if s_expr:
kwargs["s"] = self.evaluate(s_expr, selection=selection)
if c_expr:
kwargs["c"] = self.evaluate(c_expr, selection=selection)
plt.xlabel(xlabel or self.label(x))
plt.ylabel(ylabel or self.label(y))
s = plt.scatter(x_values, y_values, label=label, **kwargs)
if labels:
label_values = self.evaluate(labels, selection=selection)
for i, label_value in enumerate(label_values):
plt.annotate(label_value, (x_values[i], y_values[i]))
xerr_values = None
yerr_values = None
if cov is not None or corr is not None:
from matplotlib.patches import Ellipse
sx = self.evaluate(xerr, selection=selection)
sy = self.evaluate(yerr, selection=selection)
if corr is not None:
sxy = self.evaluate(corr, selection=selection) * sx * sy
elif cov is not None:
sxy = self.evaluate(cov, selection=selection)
cov_matrix = np.zeros((len(sx), 2, 2))
cov_matrix[:,0,0] = sx**2
cov_matrix[:,1,1] = sy**2
cov_matrix[:,0,1] = cov_matrix[:,1,0] = sxy
ax = plt.gca()
ellipse_kwargs = dict(ellipse_kwargs)
ellipse_kwargs['facecolor'] = ellipse_kwargs.get('facecolor', 'none')
ellipse_kwargs['edgecolor'] = ellipse_kwargs.get('edgecolor', 'black')
for i in range(len(sx)):
eigen_values, eigen_vectors = np.linalg.eig(cov_matrix[i])
indices = np.argsort(eigen_values)[::-1]
eigen_values = eigen_values[indices]
eigen_vectors = eigen_vectors[:,indices]
v1 = eigen_vectors[:, 0]
v2 = eigen_vectors[:, 1]
varx = cov_matrix[i, 0, 0]
vary = cov_matrix[i, 1, 1]
angle = np.arctan2(v1[1], v1[0])
# round off errors cause negative values?
if eigen_values[1] < 0 and abs((eigen_values[1]/eigen_values[0])) < 1e-10:
eigen_values[1] = 0
if eigen_values[0] < 0 or eigen_values[1] < 0:
raise ValueError('neg val')
width, height = np.sqrt(np.max(eigen_values)), np.sqrt(np.min(eigen_values))
e = Ellipse(xy=(x_values[i], y_values[i]), width=width, height=height, angle=np.degrees(angle), **ellipse_kwargs)
ax.add_artist(e)
else:
if xerr is not None:
if _issequence(xerr):
assert len(xerr) == 2, "if xerr is a sequence it should be of length 2"
xerr_values = [self.evaluate(xerr[0], selection=selection), self.evaluate(xerr[1], selection=selection)]
else:
xerr_values = self.evaluate(xerr, selection=selection)
if yerr is not None:
if _issequence(yerr):
assert len(yerr) == 2, "if yerr is a sequence it should be of length 2"
yerr_values = [self.evaluate(yerr[0], selection=selection), self.evaluate(yerr[1], selection=selection)]
else:
yerr_values = self.evaluate(yerr, selection=selection)
if xerr_values is not None or yerr_values is not None:
errorbar_kwargs = dict(errorbar_kwargs)
errorbar_kwargs['fmt'] = errorbar_kwargs.get('fmt', 'none')
plt.errorbar(x_values, y_values, yerr=yerr_values, xerr=xerr_values, **errorbar_kwargs)
return s
