def test_get_ind_mat_uniqueness(self):
"""
Tests get_ind_mat_uniqueness function using indicator matrix from the book example
"""
ind_mat = pd.DataFrame(index=range(0, 6), columns=range(0, 3))
ind_mat.loc[:, 0] = [1, 1, 1, 0, 0, 0]
ind_mat.loc[:, 1] = [0, 0, 1, 1, 0, 0]
ind_mat.loc[:, 2] = [0, 0, 0, 0, 1, 1]
ind_mat = ind_mat.values
labels_av_uniqueness = get_ind_mat_label_uniqueness(ind_mat)
first_sample_unq = labels_av_uniqueness[0]
second_sample_unq = labels_av_uniqueness[1]
third_sample_unq = labels_av_uniqueness[2]
self.assertTrue(
abs(first_sample_unq[first_sample_unq > 0].mean() -
0.8333) <= 1e-4)
self.assertTrue(
abs(second_sample_unq[second_sample_unq > 0].mean() -
0.75) <= 1e-4)
self.assertTrue(
abs(third_sample_unq[third_sample_unq > 0].mean() - 1.0) <= 1e-4)
# Test matrix av.uniqueness
self.assertTrue(
abs(labels_av_uniqueness[labels_av_uniqueness > 0].mean() -
0.8571) <= 1e-4)
def _get_synthetic_samples(ind_mat, good_samples_thresh, bad_samples_thresh):
"""
Get samples with uniqueness either > good_samples_thresh or uniqueness < bad_samples_thresh
"""
# Get mix of samples where some of them are extremely non-overlapping, the other one are highly overlapping
i = 0
unique_samples = []
for label in get_ind_mat_label_uniqueness(ind_mat):
if np.mean(label[label > 0]) > good_samples_thresh or np.mean(label[label > 0]) < bad_samples_thresh:
unique_samples.append(i)
i += 1
return unique_samples