Use n_components=3 to test actual regression

n_components and n_features were equal and one was used for the other in
some places.

sklearn/mixture/tests/test_gaussian_mixture.py

@@ -918,7 +918,7 @@ def test_property():
 
 def test_sample():
  rng = np.random.RandomState(0)
- rand_data = RandomData(rng, scale=7)
+ rand_data = RandomData(rng, scale=7, n_components=3)
  n_features, n_components = rand_data.n_features, rand_data.n_components
 
  for covar_type in COVARIANCE_TYPE:
@@ -937,7 +937,8 @@ def test_sample():
  # Just to make sure the class samples correctly
  n_samples = 20000
  X_s, y_s = gmm.sample(n_samples)
- for k in range(n_features):
+
+ for k in range(n_components):
  if covar_type == 'full':
  assert_array_almost_equal(gmm.covariances_[k],
  np.cov(X_s[y_s == k].T), decimal=1)
@@ -954,15 +955,16 @@ def test_sample():
  decimal=1)
 
  means_s = np.array([np.mean(X_s[y_s == k], 0)
- for k in range(n_features)])
+ for k in range(n_components)])
  assert_array_almost_equal(gmm.means_, means_s, decimal=1)
 
- # Check that sizes that are drawn match what is requested
- assert_equal(X_s.shape, (n_samples, n_components))
- for sample_size in range(1, 50):
- X_s, _ = gmm.sample(sample_size)
- assert_equal(X_s.shape, (sample_size, n_components))
+ # Check shapes of sampled data, see
+ # https://github.com/scikit-learn/scikit-learn/issues/7701
+ assert_equal(X_s.shape, (n_samples, n_features))
 
+ for sample_size in range(1, 100):
+ X_s, _ = gmm.sample(sample_size)
+ assert_equal(X_s.shape, (sample_size, n_features))
 
 
 @ignore_warnings(category=ConvergenceWarning)