Coverage for lasso/dimred/svd/test_clustering_betas.py: 100%

1from unittest import TestCase 

2import numpy as np 

3from lasso.dimred.svd.clustering_betas import group_betas 

4from lasso.dimred.svd.keyword_types import DetectorType, ClusterType 

5 

6 

7class TestClustering(TestCase): 

8 def test_group_betas(self): 

9 """tests correct function of the group_betas function 

10 in clustering_betas.py""" 

11 

12 fake_names = np.array(["betas_{i}".format(i=i) for i in range(25)]) 

13 fake_cluster_0 = np.random.rand(12, 3) + 5 

14 fake_cluster_1 = np.random.rand(12, 3) - 5 

15 fake_betas = np.stack([*fake_cluster_0, *fake_cluster_1, np.array([0, 0, 0])]) 

16 expected_clusters = 2 

17 expected_outliers = 1 

18 

19 # test with recommended settings 

20 beta_clusters, name_clusters = group_betas( 

21 fake_names, 

22 fake_betas, 

23 cluster=ClusterType.KMeans, 

24 detector=DetectorType.LocalOutlierFactor, 

25 cluster_params=dict( 

26 n_clusters=expected_clusters, 

27 ), 

28 ) 

29 

30 # verify correct type of output 

31 self.assertIsInstance(beta_clusters, list) 

32 self.assertIsInstance(name_clusters, list) 

33 

34 # verify that beta_clusters and name_clusters correspond to each other 

35 self.assertEqual(len(beta_clusters), len(name_clusters)) 

36 # verify that beta_clusters contains as many clusters as searched for 

37 # inkluding one outlier 

38 self.assertEqual(len(beta_clusters), expected_clusters + expected_outliers) 

39 

40 # verify that entries correspond to each other 

41 for c, cluster in enumerate(name_clusters): 

42 for e, entry in enumerate(cluster): 

43 index = np.where(fake_names == entry)[0] 

44 self.assertTrue((fake_betas[index] - beta_clusters[c][e]).max() == 0) 

45 

46 # verify differen keyword kombinations 

47 

48 for cluster_type in ClusterType.get_cluster_type_name(): 

49 for detector_type in DetectorType.get_detector_type_name(): 

50 beta_clusters, name_clusters = group_betas( 

51 fake_names, fake_betas, cluster=cluster_type, detector=detector_type 

52 ) 

53 

54 # verify correct output 

55 self.assertIsInstance(beta_clusters, list) 

56 self.assertIsInstance(name_clusters, list) 

57 self.assertEqual(len(beta_clusters), len(name_clusters))