-
Notifications
You must be signed in to change notification settings - Fork 45
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
1 changed file
with
73 additions
and
0 deletions.
There are no files selected for viewing
73 changes: 73 additions & 0 deletions
73
tests/integration/reports/single_table/_properties/test_data_validity.py
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,73 @@ | ||
import pandas as pd | ||
|
||
from sdmetrics.demos import load_demo | ||
from sdmetrics.reports.single_table._properties import DataValidity | ||
|
||
|
||
class TestDataValidity: | ||
|
||
def test_get_score(self): | ||
"""Test the ``get_score`` method""" | ||
# Setup | ||
real_data, synthetic_data, metadata = load_demo('single_table') | ||
|
||
# Run | ||
data_validity_property = DataValidity() | ||
score = data_validity_property.get_score(real_data, synthetic_data, metadata) | ||
|
||
# Assert | ||
expected_details_dict = { | ||
'Column': [ | ||
'start_date', 'end_date', 'salary', 'duration', 'student_id', | ||
'high_perc', 'high_spec', 'mba_spec', 'second_perc', 'gender', | ||
'degree_perc', 'placed', 'experience_years', 'employability_perc', | ||
'mba_perc', 'work_experience', 'degree_type' | ||
], | ||
'Metric': [ | ||
'BoundaryAdherence', 'BoundaryAdherence', 'BoundaryAdherence', 'BoundaryAdherence', | ||
'KeyUniqueness', 'BoundaryAdherence', 'CategoryAdherence', 'CategoryAdherence', | ||
'BoundaryAdherence', 'CategoryAdherence', 'BoundaryAdherence', 'CategoryAdherence', | ||
'BoundaryAdherence', 'BoundaryAdherence', 'BoundaryAdherence', 'CategoryAdherence', | ||
'CategoryAdherence' | ||
], | ||
'Score': [ | ||
0.8503937007874016, 0.8615384615384616, 0.9444444444444444, | ||
1.0, 1.0, 0.8651162790697674, 1.0, 1.0, 0.9255813953488372, | ||
1.0, 0.9441860465116279, 1.0, 1.0, 0.8883720930232558, | ||
0.8930232558139535, 1.0, 1.0 | ||
] | ||
} | ||
expected_details = pd.DataFrame(expected_details_dict) | ||
pd.testing.assert_frame_equal(data_validity_property.details, expected_details) | ||
assert score == 0.9513326868551618 | ||
|
||
def test_get_score_errors(self): | ||
"""Test the ``get_score`` method when the metrics are raising errors for some columns.""" | ||
# Setup | ||
real_data, synthetic_data, metadata = load_demo('single_table') | ||
|
||
real_data['start_date'].iloc[0] = 0 | ||
real_data['employability_perc'].iloc[2] = 'a' | ||
|
||
# Run | ||
data_validity_property = DataValidity() | ||
|
||
expected_message_1 = ( | ||
"TypeError: '<=' not supported between instances of 'int' and 'Timestamp'" | ||
) | ||
expected_message_2 = ( | ||
"TypeError: '<=' not supported between instances of 'float' and 'str'" | ||
) | ||
|
||
score = data_validity_property.get_score(real_data, synthetic_data, metadata) | ||
|
||
# Assert | ||
|
||
details = data_validity_property.details | ||
details_nan = details.loc[pd.isna(details['Score'])] | ||
column_names_nan = details_nan['Column'].tolist() | ||
error_messages = details_nan['Error'].tolist() | ||
assert column_names_nan == ['start_date', 'employability_perc'] | ||
assert error_messages[0] == expected_message_1 | ||
assert error_messages[1] == expected_message_2 | ||
assert score == 0.9622593255151395 |