class TestCAResults(object):
def setup(self):
"""Data from table 9.11 in Legendre & Legendre 1998."""
self.X = np.loadtxt(get_data_path('L&L_CA_data'))
self.ordination = CA(self.X, ['Site1', 'Site2', 'Site3'],
['Species1', 'Species2', 'Species3'])
def test_scaling2(self):
scores = self.ordination.scores(scaling=2)
# p. 460 L&L 1998
F_hat = np.array([[0.40887, -0.06955],
[-0.11539, 0.29977],
[-0.30997, -0.18739]])
npt.assert_almost_equal(*normalize_signs(F_hat, scores.species),
decimal=5)
V_hat = np.array([[-0.84896, -0.88276],
[-0.22046, 1.34482],
[1.66697, -0.47032]])
npt.assert_almost_equal(*normalize_signs(V_hat, scores.site),
decimal=5)
def test_scaling1(self):
scores = self.ordination.scores(scaling=1)
# p. 458
V = np.array([[1.31871, -0.34374],
[-0.37215, 1.48150],
[-0.99972, -0.92612]])
npt.assert_almost_equal(*normalize_signs(V, scores.species), decimal=5)
F = np.array([[-0.26322, -0.17862],
[-0.06835, 0.27211],
[0.51685, -0.09517]])
npt.assert_almost_equal(*normalize_signs(F, scores.site), decimal=5)
def test_maintain_chi_square_distance_scaling1(self):
"""In scaling 1, chi^2 distance among rows (sites) is equal to
euclidean distance between them in transformed space."""
frequencies = self.X / self.X.sum()
chi2_distances = chi_square_distance(frequencies)
transformed_sites = self.ordination.scores(1).site
euclidean_distances = pdist(transformed_sites, 'euclidean')
npt.assert_almost_equal(chi2_distances, euclidean_distances)
def test_maintain_chi_square_distance_scaling2(self):
"""In scaling 2, chi^2 distance among columns (species) is
equal to euclidean distance between them in transformed space."""
frequencies = self.X / self.X.sum()
chi2_distances = chi_square_distance(frequencies, between_rows=False)
transformed_species = self.ordination.scores(2).species
euclidean_distances = pdist(transformed_species, 'euclidean')
npt.assert_almost_equal(chi2_distances, euclidean_distances)
class TestCAResults(object):
def setup(self):
"""Data from table 9.11 in Legendre & Legendre 1998."""
self.X = np.loadtxt(get_data_path('L&L_CA_data'))
self.ordination = CA(self.X, ['Site1', 'Site2', 'Site3'],
['Species1', 'Species2', 'Species3'])
def test_scaling2(self):
scores = self.ordination.scores(scaling=2)
# p. 460 L&L 1998
F_hat = np.array([[0.40887, -0.06955], [-0.11539, 0.29977],
[-0.30997, -0.18739]])
npt.assert_almost_equal(*normalize_signs(F_hat, scores.species),
decimal=5)
V_hat = np.array([[-0.84896, -0.88276], [-0.22046, 1.34482],
[1.66697, -0.47032]])
npt.assert_almost_equal(*normalize_signs(V_hat, scores.site),
decimal=5)
def test_scaling1(self):
scores = self.ordination.scores(scaling=1)
# p. 458
V = np.array([[1.31871, -0.34374], [-0.37215, 1.48150],
[-0.99972, -0.92612]])
npt.assert_almost_equal(*normalize_signs(V, scores.species), decimal=5)
F = np.array([[-0.26322, -0.17862], [-0.06835, 0.27211],
[0.51685, -0.09517]])
npt.assert_almost_equal(*normalize_signs(F, scores.site), decimal=5)
def test_maintain_chi_square_distance_scaling1(self):
"""In scaling 1, chi^2 distance among rows (sites) is equal to
euclidean distance between them in transformed space."""
frequencies = self.X / self.X.sum()
chi2_distances = chi_square_distance(frequencies)
transformed_sites = self.ordination.scores(1).site
euclidean_distances = pdist(transformed_sites, 'euclidean')
npt.assert_almost_equal(chi2_distances, euclidean_distances)
def test_maintain_chi_square_distance_scaling2(self):
"""In scaling 2, chi^2 distance among columns (species) is
equal to euclidean distance between them in transformed space."""
frequencies = self.X / self.X.sum()
chi2_distances = chi_square_distance(frequencies, between_rows=False)
transformed_species = self.ordination.scores(2).species
euclidean_distances = pdist(transformed_species, 'euclidean')
npt.assert_almost_equal(chi2_distances, euclidean_distances)
def setup(self):
"""Data from table 9.11 in Legendre & Legendre 1998."""
self.X = np.loadtxt(get_data_path('L&L_CA_data'))
self.ordination = CA(self.X, ['Site1', 'Site2', 'Site3'],
['Species1', 'Species2', 'Species3'])
def test_negative(self):
X = np.array([[1, 2], [-0.1, -2]])
with npt.assert_raises(ValueError):
CA(X, None, None)
def setup(self):
"""Data from table 9.11 in Legendre & Legendre 1998."""
self.X = np.loadtxt(get_data_path('L&L_CA_data'))
self.ordination = CA(self.X, ['Site1', 'Site2', 'Site3'],
['Species1', 'Species2', 'Species3'])
import os
import numpy as np
from skbio.math.stats.ordination import CA, RDA, CCA
path = os.path.dirname(os.path.abspath(__file__))
def get_path(fn):
return os.path.join(path, os.pardir, 'math', 'stats', 'ordination',
'test', 'data', fn)
X = np.loadtxt(get_path('L&L_CA_data'))
ordint = CA(X)
ordint.biplot(1)
ordint.biplot(2)
Y = np.loadtxt(get_path('example2_Y'))
X = np.loadtxt(get_path('example2_X')).reshape(-1, 4, order='F')
ordint = RDA(Y, X)
ordint.biplot()
Y = np.loadtxt(get_path('example3_Y'))
X = np.loadtxt(get_path('example3_X')).reshape(-1, 4, order='F')
ordint = CCA(Y, X)
ordint.biplot()
