Use pytest.approx in unit tests

tests/test_beta_oddsratio.py

@@ -14,6 +14,8 @@
 #   You should have received a copy of the GNU Affero General Public License
 #   along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
+from pytest import approx
+
 import numpy as np
 from scipy import stats
 
@@ -38,7 +40,7 @@ def test_beta_ratio_cdf_vs_empirical():
 	y2 = [(sample < xx).sum()/sample.size for xx in x]
 	
 	# Allow 0.01 tolerance
-	assert (np.abs(y1 - y2) < 0.01).all()
+	assert y1 == approx(y2, abs=0.01)
 
 def test_beta_ratio_ppf_vs_empirical():
 	"""Compare beta_ratio.ppf with empirical distribution"""
@@ -59,7 +61,7 @@ def test_beta_ratio_ppf_vs_empirical():
 	y2 = np.quantile(sample, x)
 	
 	# Allow 0.01 tolerance
-	assert (np.abs(y1 - y2) < 0.01).all()
+	assert y1 == approx(y2, abs=0.01)
 
 def test_beta_ratio_mean_vs_empirical():
 	"""Compare beta_ratio.mean (vs _munp) with empirical mean"""
@@ -75,7 +77,7 @@ def test_beta_ratio_mean_vs_empirical():
 	sample = (samples_p1 / (1 - samples_p1)) / (samples_p2 / (1 - samples_p2))
 	
 	# Allow 0.01 tolerance
-	assert np.abs(dist.mean() - sample.mean()) < 0.01
+	assert dist.mean() == approx(sample.mean(), abs=0.01)
 
 def test_beta_ratio_var_vs_empirical():
 	"""Compare beta_ratio.var (vs _munp) with empirical variance"""
@@ -91,4 +93,4 @@ def test_beta_ratio_var_vs_empirical():
 	sample = (samples_p1 / (1 - samples_p1)) / (samples_p2 / (1 - samples_p2))
 	
 	# Allow 0.01 tolerance
-	assert np.abs(dist.var() - sample.var()) < 0.01
+	assert dist.var() == approx(sample.var(), abs=0.01)

tests/test_beta_ratio.py

@@ -14,6 +14,8 @@
 #   You should have received a copy of the GNU Affero General Public License
 #   along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
+from pytest import approx
+
 import numpy as np
 from scipy import stats
 
@@ -32,10 +34,7 @@ def test_beta_ratio_vs_jsaffer_pdf():
 	
 	# Compare with expected values from jsaffer implementation
 	expected = np.load('tests/beta_ratio_vs_jsaffer.npy', allow_pickle=False)[0]
-	
-	# Allow 1e-10 tolerance
-	diff = np.abs(y - expected)
-	assert (diff < 1e-10).all()
+	assert y == approx(expected)
 
 def test_beta_ratio_vs_jsaffer_cdf():
 	"""Compare beta_ratio.cdf with result from https://github.com/jsaffer/beta_quotient_distribution"""
@@ -50,10 +49,7 @@ def test_beta_ratio_vs_jsaffer_cdf():
 	
 	# Compare with expected values from jsaffer implementation
 	expected = np.load('tests/beta_ratio_vs_jsaffer.npy', allow_pickle=False)[1]
-	
-	# Allow 1e-10 tolerance
-	diff = np.abs(y - expected)
-	assert (diff < 1e-10).all()
+	assert y == approx(expected)
 
 def _gen_beta_ratio_vs_jsaffer():
 	"""Generate beta_ratio_vs_jsaffer.npy for test_beta_ratio_vs_jsaffer_pdf/cdf"""
@@ -82,7 +78,7 @@ def test_beta_ratio_mean_vs_empirical():
 	sample = samples_p1 / samples_p2
 	
 	# Allow 0.01 tolerance
-	assert np.abs(dist.mean() - sample.mean()) < 0.01
+	assert dist.mean() == approx(sample.mean(), abs=0.01)
 
 def test_beta_ratio_var_vs_empirical():
 	"""Compare beta_ratio.var (via beta_ratio._munp) with empirical variance"""
@@ -98,4 +94,4 @@ def test_beta_ratio_var_vs_empirical():
 	sample = samples_p1 / samples_p2
 	
 	# Allow 0.01 tolerance
-	assert np.abs(dist.var() - sample.var()) < 0.01
+	assert dist.var() == approx(sample.var(), abs=0.01)

tests/test_hdi.py

@@ -14,11 +14,13 @@
 #   You should have received a copy of the GNU Affero General Public License
 #   along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-import yli
+from pytest import approx
 
 import numpy as np
 from scipy import stats
 
+import yli
+
 def test_hdi_beta_vs_r():
 	"""Compare yli.hdi for beta distribution with R binom.bayes"""
 	
@@ -37,6 +39,4 @@ def test_hdi_beta_vs_r():
 	#1  bayes 12 250     13    239 0.0515873 0.02594348 0.0793006 0.05
 	expected = np.array([0.02594348, 0.0793006])
 	
-	# Allow 1e-5 tolerance
-	diff = np.abs(np.array(hdi) - expected)
-	assert (diff < 1e-5).all()
+	assert hdi == approx(expected)