Baunsgaard commented on a change in pull request #892:
URL: https://github.com/apache/systemml/pull/892#discussion_r410382542
##########
File path: src/main/python/tests/test_matrix_aggregations.py
##########
@@ -84,6 +86,63 @@ def test_var2(self):
def test_var3(self):
self.assertTrue(np.allclose(sds.matrix(m1).var(axis=1).compute(),
m1.var(axis=1, ddof=1).reshape(dim, 1)))
+ def test_rand_basic(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+ sparsity = 0.2
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed, sparsity=sparsity).compute()
+
+ self.assertTrue(m.shape == shape)
+ self.assertTrue((m.min() >= min_max[0]) and (m.max() <= min_max[1]))
+
+ # sparsity
+ m_flat = m.flatten('F')
+ count, bins, patches = plt.hist(m_flat)
+
+ non_zero_value_percent = sum(count[1:]) * 100 / count[0]
+ e = 0.05
+ self.assertTrue((non_zero_value_percent >= (sparsity - e) * 100)
+ and (non_zero_value_percent <= (sparsity + e) * 100))
+ self.assertTrue(sum(count) == (shape[0] * shape[1]))
+
+ def test_rand_distribution(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m_flat = m.flatten('F')
+
+ dist = best_distribution(m_flat)
+ self.assertTrue(dist == 'uniform')
+
+ m1 = sds.rand(rows=shape[0], cols=shape[1], pdf="normal",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m1_flat = m1.flatten('F')
+
+ dist = best_distribution(m1_flat)
+ self.assertTrue(dist == 'norm')
+
+
Review comment:
(OCD comment) double newline
##########
File path: src/main/python/tests/test_matrix_aggregations.py
##########
@@ -84,6 +86,63 @@ def test_var2(self):
def test_var3(self):
self.assertTrue(np.allclose(sds.matrix(m1).var(axis=1).compute(),
m1.var(axis=1, ddof=1).reshape(dim, 1)))
+ def test_rand_basic(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+ sparsity = 0.2
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed, sparsity=sparsity).compute()
+
+ self.assertTrue(m.shape == shape)
+ self.assertTrue((m.min() >= min_max[0]) and (m.max() <= min_max[1]))
+
+ # sparsity
+ m_flat = m.flatten('F')
+ count, bins, patches = plt.hist(m_flat)
+
+ non_zero_value_percent = sum(count[1:]) * 100 / count[0]
+ e = 0.05
+ self.assertTrue((non_zero_value_percent >= (sparsity - e) * 100)
+ and (non_zero_value_percent <= (sparsity + e) * 100))
+ self.assertTrue(sum(count) == (shape[0] * shape[1]))
+
+ def test_rand_distribution(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m_flat = m.flatten('F')
+
+ dist = best_distribution(m_flat)
+ self.assertTrue(dist == 'uniform')
+
+ m1 = sds.rand(rows=shape[0], cols=shape[1], pdf="normal",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m1_flat = m1.flatten('F')
+
+ dist = best_distribution(m1_flat)
+ self.assertTrue(dist == 'norm')
+
+
+def best_distribution(data):
+ distributions = ['norm', 'uniform']
+ result = dict()
+
+ for dist in distributions:
+ param = getattr(st, dist).fit(data)
+
+ D, p_value = st.kstest(data, dist, args=param)
+ result[dist] = p_value
+
+ best_dist = max(result, key=result.get)
+ return best_dist
+
Review comment:
if it removes the newlines, that would be great :+1:
##########
File path: src/main/python/tests/test_matrix_aggregations.py
##########
@@ -84,6 +86,63 @@ def test_var2(self):
def test_var3(self):
self.assertTrue(np.allclose(sds.matrix(m1).var(axis=1).compute(),
m1.var(axis=1, ddof=1).reshape(dim, 1)))
+ def test_rand_basic(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+ sparsity = 0.2
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed, sparsity=sparsity).compute()
+
+ self.assertTrue(m.shape == shape)
+ self.assertTrue((m.min() >= min_max[0]) and (m.max() <= min_max[1]))
+
+ # sparsity
+ m_flat = m.flatten('F')
+ count, bins, patches = plt.hist(m_flat)
+
+ non_zero_value_percent = sum(count[1:]) * 100 / count[0]
+ e = 0.05
+ self.assertTrue((non_zero_value_percent >= (sparsity - e) * 100)
+ and (non_zero_value_percent <= (sparsity + e) * 100))
+ self.assertTrue(sum(count) == (shape[0] * shape[1]))
+
+ def test_rand_distribution(self):
+ seed = 15
+ shape = (20, 20)
+ min_max = (0, 1)
+
+ m = sds.rand(rows=shape[0], cols=shape[1], pdf="uniform",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m_flat = m.flatten('F')
+
+ dist = best_distribution(m_flat)
+ self.assertTrue(dist == 'uniform')
+
+ m1 = sds.rand(rows=shape[0], cols=shape[1], pdf="normal",
min=min_max[0], max=min_max[1],
+ seed=seed).compute()
+
+ m1_flat = m1.flatten('F')
+
+ dist = best_distribution(m1_flat)
+ self.assertTrue(dist == 'norm')
+
+
+def best_distribution(data):
+ distributions = ['norm', 'uniform']
+ result = dict()
+
+ for dist in distributions:
+ param = getattr(st, dist).fit(data)
+
+ D, p_value = st.kstest(data, dist, args=param)
+ result[dist] = p_value
+
+ best_dist = max(result, key=result.get)
+ return best_dist
+
Review comment:
(OCD comment) double newline
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
