def test_transform(self):
for tfm in [None, ILRTransform, ALRTransform]:
with self.subTest(tfm=tfm):
if callable(tfm):
T = tfm()
comp = T.transform(self.comp3d)
transform = T.inverse_transform
else:
transform = None
comp = self.comp2d
plot_pca_vectors(comp, transform=transform)
comp.loc[:, chem].pyroplot.scatter(ax=ax[0], c=t10b3[ix], **kwargs)
plt.show()
#######################################################################################
# We can take the mean and covariance in log-space to create covariance ellipses and
# vectors using principal component analysis:
#
kwargs = dict(ax=ax[1], transform=from_log, nstds=3)
ax[1].set_title("Covariance Ellipses and PCA Vectors")
for ix, sample in enumerate(df.Sample.unique()):
comp = df.query("Sample == {}".format(sample))
tcomp = to_log(comp.loc[:, chem])
plot_stdev_ellipses(tcomp.values,
color=t10b3[ix],
resolution=1000,
**kwargs)
plot_pca_vectors(tcomp.values, ls="-", lw=0.5, color="k", **kwargs)
plt.show()
#######################################################################################
# We can also look at data density (here using kernel density estimation)
# in logratio-space:
#
kwargs = dict(ax=ax[-2], bins=100, axlabels=False)
ax[-2].set_title("Individual Density, with Contours")
for ix, sample in enumerate(df.Sample.unique()):
comp = df.query("Sample == {}".format(sample))
comp.loc[:, chem].pyroplot.density(cmap="Blues", vmin=0.05, **kwargs)
comp.loc[:, chem].pyroplot.density(
contours=[0.68, 0.95],
cmap="Blues_r",
contour_labels={
def test_default(self):
for comp in [self.comp2d, self.comp3d]:
with self.subTest(comp=comp):
plot_pca_vectors(comp)
def test_axis_specified(self):
for comp in [self.comp2d, self.comp3d]:
with self.subTest(comp=comp):
fig, ax = plt.subplots()
plot_pca_vectors(comp, ax=ax)