"""! @brief Unit-tests for EMA algorithm. @authors Andrei Novikov (pyclustering@yandex.ru) @date 2014-2020 @copyright BSD-3-Clause """ import unittest # Generate images without having a window appear. import matplotlib matplotlib.use('Agg') from pyclustering.cluster.ema import ema, ema_observer, ema_initializer, ema_init_type, ema_visualizer from pyclustering.utils import read_sample from pyclustering.samples.definitions import SIMPLE_SAMPLES, FCPS_SAMPLES, FAMOUS_SAMPLES class EmaUnitTest(unittest.TestCase): def templateDataClustering(self, sample_path, amount_clusters, expected_clusters_sizes, init_type = ema_init_type.KMEANS_INITIALIZATION): testing_result = False attempts = 10 for _ in range(attempts): sample = read_sample(sample_path) means, variances = None, None if init_type is not ema_init_type.KMEANS_INITIALIZATION: means, variances = ema_initializer(sample, amount_clusters).initialize(init_type) ema_instance = ema(sample, amount_clusters, means, variances) ema_instance.process() clusters = ema_instance.get_clusters() centers = ema_instance.get_centers() covariances = ema_instance.get_covariances() probabilities = ema_instance.get_probabilities() assert len(centers) == len(clusters) assert len(covariances) == len(clusters) assert len(probabilities) == len(clusters) for cluster_probability in probabilities: assert len(cluster_probability) == len(sample) for index_point in range(len(sample)): total_probability = 0.0 for cluster_probability in probabilities: total_probability += cluster_probability[index_point] assert abs(total_probability - 1.0) <= 0.00001 obtained_cluster_sizes = [len(cluster) for cluster in clusters] if len(sample) != sum(obtained_cluster_sizes): continue if expected_clusters_sizes is not None: obtained_cluster_sizes.sort() expected_clusters_sizes.sort() if obtained_cluster_sizes != expected_clusters_sizes: continue testing_result = True break assert testing_result is True def testClusteringSampleSimple01(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 2, [5, 5]) def testClusteringSampleSimple01RandomInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 2, [5, 5], ema_init_type.RANDOM_INITIALIZATION) def testClusteringSampleSimple01OneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 1, [10]) def testClusteringSampleSimple01OneClusterRandomInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 1, [10], ema_init_type.RANDOM_INITIALIZATION) def testClusteringSampleSimple01ThreeCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 3, None) def testClusteringSampleSimple01ThreeClusterRandomInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 3, None, ema_init_type.RANDOM_INITIALIZATION) def testClusteringSampleSimple01TenCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 10, None) def testClusteringSampleSimple01SevenCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 7, None) def testClusteringSampleSimple02(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE2, 3, [5, 8, 10]) def testClusteringSampleSimple02OneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE2, 1, [23]) def testClusteringSampleSimple02ThreeCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 3, None) def testClusteringSampleSimple03(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE3, 4, [10, 10, 10, 30]) def testClusteringSampleSimple03OneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE3, 1, [60]) def testClusteringSampleSimple05(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE5, 4, [15, 15, 15, 15]) def testClusteringSampleSimple05OneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE5, 1, [60]) def testClusteringCommonOldFaithful(self): self.templateDataClustering(FAMOUS_SAMPLES.SAMPLE_OLD_FAITHFUL, 2, [97, 175]) def testClusteringFcpsLsun(self): self.templateDataClustering(FCPS_SAMPLES.SAMPLE_LSUN, 3, [100, 101, 202]) def testClusteringOneDimensionalData(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE7, 2, [10, 10]) def testClusteringOneDimensionalDataRandomInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE7, 2, [10, 10], ema_init_type.RANDOM_INITIALIZATION) def testClusteringOneDimensionalDataOneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE7, 1, [20]) def testClusteringThreeDimensionalData(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE11, 2, [10, 10]) def testClusteringThreeDimensionalDataOneCluster(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE11, 1, [20]) def testClusteringTotallySimilarObjects(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE12, 1, None) def testClusteringTotallySimilarObjectsTwoClusters(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE12, 2, None) def testClusteringTotallySimilarObjectsThreeClustersRandonInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE12, 3, None, ema_init_type.RANDOM_INITIALIZATION) def testClusteringTotallySimilarObjectsFiveClusters(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE12, 5, None) def testClusteringTotallySimilarObjectsFiveClustersRandomInit(self): self.templateDataClustering(SIMPLE_SAMPLES.SAMPLE_SIMPLE12, 5, None, ema_init_type.RANDOM_INITIALIZATION) def testObserver(self): sample = read_sample(SIMPLE_SAMPLES.SAMPLE_SIMPLE2) means, variances = ema_initializer(sample, 3).initialize(ema_init_type.RANDOM_INITIALIZATION) observer_instance = ema_observer() ema_instance = ema(sample, 3, means, variances, observer_instance) ema_instance.process() observer_length = len(observer_instance) assert observer_length > 0 assert observer_length == len(observer_instance.get_evolution_clusters()) assert observer_length == len(observer_instance.get_evolution_covariances()) assert observer_length == len(observer_instance.get_evolution_means()) assert observer_length == observer_instance.get_iterations() def testVisualizerNoFailures(self): sample = read_sample(SIMPLE_SAMPLES.SAMPLE_SIMPLE3) means, variances = ema_initializer(sample, 4).initialize(ema_init_type.RANDOM_INITIALIZATION) observer_instance = ema_observer() ema_instance = ema(sample, 4, means, variances, observer_instance) ema_instance.process() clusters = ema_instance.get_clusters() means = ema_instance.get_centers() covariances = ema_instance.get_covariances() ema_visualizer.show_clusters(clusters, sample, covariances, means) ema_visualizer.animate_cluster_allocation(sample, observer_instance) def test_incorrect_data(self): self.assertRaises(ValueError, ema, [], 2) def test_incorrect_amount_clusters(self): self.assertRaises(ValueError, ema, [[0], [1], [2]], 0)