def setUp(self):
dates = pd.date_range(start='2018.01.01', end='2018.01.03', freq='1h')
np.random.seed(42)
self.df = pd.DataFrame({
'date_column':
list(map(str, dates)),
"outcome":
np.random.choice(['Noticeable', 'Totally fine', 'Distracting'],
size=len(dates)),
"intervention":
np.random.choice(['A', 'B', ''], p=[.1, .1, .8], size=len(dates)),
})
self.pdf = PrePostDF()
def test_fisher_test(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.pps['A'] = self.pdf._prepost_generator('A')
ft = self.pdf.fisher_test('A')
self.assertListEqual(ft.columns.tolist(),
['Noticeable', 'Totally fine', 'ratio'])
self.assertSequenceEqual(ft.shape, (2, 3))
self.assertTrue((ft >= 0).all().all())
self.assertTrue(ft.notnull().all().all())
def test_basic_info(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
capturedOutput = StringIO() # Create StringIO object
sys.stdout = capturedOutput # and redirect stdout.
self.pdf.basic_info()
sys.stdout = sys.__stdout__ # Reset redirect.
self.assertTrue(
'Latest recording: 2018.01.03' in capturedOutput.getvalue())
self.assertTrue(
'Latest recording: 2018.01.03' in capturedOutput.getvalue())
def test_p_distribution(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.outcomes(positive_outcomes=['Totally fine'],
negative_outcomes=['Noticeable', 'Distracting'])
posterior = self.pdf._p_distribution(intervention='A')
# Broad test to confirm function outut has not materially changed since last manual review
self.assertListEqual(
posterior.round(1).tolist(),
[0.0, 0.1, 0.2, 0.2, 0.2, 0.1, 0.1, 0.0, 0.0, 0.0])
def test_outcomes(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.outcomes(positive_outcomes=['Totally fine'],
negative_outcomes=['Noticeable', 'Distracting'])
# Broad test to confirm function outut has not materially changed since last manual review
self.assertListEqual(self.pdf.confusion_matrix.loc[''].tolist(),
[4, 9, 8, 0, 14])
self.assertListEqual(self.pdf.confusion_matrix.loc['A'].tolist(),
[1, 3, 0, 1, 3])
self.assertListEqual(self.pdf.confusion_matrix.loc['B'].tolist(),
[1, 0, 2, 0, 3])
def test_prepost_generator(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
pps = self.pdf._prepost_generator('A')
self.assertListEqual(pps.columns.tolist(), [('pre', 'Distracting'),
('pre', 'Noticeable'),
('pre', 'Totally fine'),
('post', 'Distracting'),
('post', 'Noticeable'),
('post', 'Totally fine')])
self.assertSequenceEqual(pps.shape, (8, 6))
self.assertTrue((pps >= 0).all().all())
self.assertTrue(pps.notnull().all().all())
class TestPrePostDF(unittest.TestCase):
def setUp(self):
dates = pd.date_range(start='2018.01.01', end='2018.01.03', freq='1h')
np.random.seed(42)
self.df = pd.DataFrame({
'date_column':
list(map(str, dates)),
"outcome":
np.random.choice(['Noticeable', 'Totally fine', 'Distracting'],
size=len(dates)),
"intervention":
np.random.choice(['A', 'B', ''], p=[.1, .1, .8], size=len(dates)),
})
self.pdf = PrePostDF()
def test_import_reporter(self):
output = PrePostDF.import_reporter(self.df)
self.assertEqual(output.index.size, 49)
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_basic_info(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
capturedOutput = StringIO() # Create StringIO object
sys.stdout = capturedOutput # and redirect stdout.
self.pdf.basic_info()
sys.stdout = sys.__stdout__ # Reset redirect.
self.assertTrue(
'Latest recording: 2018.01.03' in capturedOutput.getvalue())
self.assertTrue(
'Latest recording: 2018.01.03' in capturedOutput.getvalue())
@patch('pandas.read_csv')
def test_read_csv(self, mock_csv):
mock_csv.return_value = self.df.rename(
columns={'intervention': 'Choose intervention'})
output = self.pdf.read_csv('filename')
self.assertEqual(output.index.size, 49)
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_add_outcome(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_outcome(filename='data/reporter20180729')
self.assertEqual(self.pdf.df.index.size, 49)
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_add_intervention(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.assertEqual(self.pdf.df.index.size, 49)
self.assertEqual(list(self.pdf.pps.keys()), ['', 'A', 'B'])
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_prepost_generator(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
pps = self.pdf._prepost_generator('A')
self.assertListEqual(pps.columns.tolist(), [('pre', 'Distracting'),
('pre', 'Noticeable'),
('pre', 'Totally fine'),
('post', 'Distracting'),
('post', 'Noticeable'),
('post', 'Totally fine')])
self.assertSequenceEqual(pps.shape, (8, 6))
self.assertTrue((pps >= 0).all().all())
self.assertTrue(pps.notnull().all().all())
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_fisher_test(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.pps['A'] = self.pdf._prepost_generator('A')
ft = self.pdf.fisher_test('A')
self.assertListEqual(ft.columns.tolist(),
['Noticeable', 'Totally fine', 'ratio'])
self.assertSequenceEqual(ft.shape, (2, 3))
self.assertTrue((ft >= 0).all().all())
self.assertTrue(ft.notnull().all().all())
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_outcomes(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.outcomes(positive_outcomes=['Totally fine'],
negative_outcomes=['Noticeable', 'Distracting'])
# Broad test to confirm function outut has not materially changed since last manual review
self.assertListEqual(self.pdf.confusion_matrix.loc[''].tolist(),
[4, 9, 8, 0, 14])
self.assertListEqual(self.pdf.confusion_matrix.loc['A'].tolist(),
[1, 3, 0, 1, 3])
self.assertListEqual(self.pdf.confusion_matrix.loc['B'].tolist(),
[1, 0, 2, 0, 3])
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_p_distribution(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.outcomes(positive_outcomes=['Totally fine'],
negative_outcomes=['Noticeable', 'Distracting'])
posterior = self.pdf._p_distribution(intervention='A')
# Broad test to confirm function outut has not materially changed since last manual review
self.assertListEqual(
posterior.round(1).tolist(),
[0.0, 0.1, 0.2, 0.2, 0.2, 0.1, 0.1, 0.0, 0.0, 0.0])
@patch('preposterous.preposterous.PrePostDF.read_csv')
def test_calculate_relative_effectiveness(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.pdf.add_outcome(filename='data/reporter20180729')
self.pdf.outcomes(positive_outcomes=['Totally fine'],
negative_outcomes=['Noticeable', 'Distracting'])
self.pdf.calculate_relative_effectiveness(interventions=['A', 'B'])
# Broad test to confirm function outut has not materially changed since last manual review
self.assertListEqual(
self.pdf.relative_effectiveness['A'].round(1).tolist(), [
0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.1, 0.2, 0.2, 0.2, 0.2, 0.1,
0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0
])
@patch('matplotlib.pyplot.Figure.savefig')
def test_plot_relative_effectiveness(self, savefig_patch):
self.pdf.relative_effectiveness = pd.DataFrame({'A': [0, .1, .9]},
index=[.25, .5, .75])
fig = self.pdf.plot_relative_effectiveness()
# General test to confirm plot was created, and would have been saved to disk
savefig_patch.assert_called()
def test_add_intervention(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_intervention(filename='data/reporter20180729')
self.assertEqual(self.pdf.df.index.size, 49)
self.assertEqual(list(self.pdf.pps.keys()), ['', 'A', 'B'])
def test_add_outcome(self, mock_csv):
mock_csv.return_value = PrePostDF.import_reporter(self.df)
self.pdf.add_outcome(filename='data/reporter20180729')
self.assertEqual(self.pdf.df.index.size, 49)
def test_import_reporter(self):
output = PrePostDF.import_reporter(self.df)
self.assertEqual(output.index.size, 49)