def test_model_prediction_differential(save_file='test_data_predictions.csv'):
"""
This test compares the prediction result similarity of the
current model with the previous model's results.
"""
# given
previous_model_df = load_dataset('test_data_predictions.csv')
previous_model_predictions = previous_model_df.predictions.values
data = load_dataset(file_name='lending_data.csv')
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
# take a slice with no input validation issues
multiple_test_json = X_validation[99:600]
# when
response = make_prediction(input_data=multiple_test_json)
current_model_predictions = response.get('predictions')
# then
# difference the current model vs. the old model
assert len(previous_model_predictions) == len(current_model_predictions)
# perform differential test
for previous_value, current_value in zip(previous_model_predictions,
current_model_predictions):
# convert numpy float to Python float
previous_value = previous_value.item()
current_value = current_value.item()
assert math.isclose(previous_value,
current_value,
rel_tol=config.ACCEPTABLE_MODEL_DIFFERENCE)
def test_make_multiple_predictions():
"""Make multiple predictions on validation data"""
# given
data = load_dataset(file_name='lending_data.csv')
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
multiple_validation_json = X_validation.to_json(orient='records')
# when
subject = make_prediction(input_data=multiple_validation_json)
# then
assert subject is not None
assert len(subject.get('predictions')) == 44358
def test_prediction_endpoint_validation_200(flask_test_client):
# given
# get validation data from lending_model package
data = load_dataset(file_name=config.TRAINING_DATA_FILE)
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
post_json = X_validation.to_json(orient='records')
# when
response = flask_test_client.post('/v1/predict/lending_rate',
json=post_json)
# then
assert response.status_code == 200
def test_make_single_prediction():
"""Make a single prediction on validation data."""
# given
data = load_dataset(file_name='lending_data.csv')
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
single_validation_json = X_validation[0:1].to_json(orient='records')
# when
subject = make_prediction(input_data=single_validation_json)
# then
assert subject is not None
assert isinstance(subject.get('predictions')[0], float)
assert subject.get('predictions')[0] > 0
def capture_predictions(save_file='test_data_predictions.csv'):
"""Save test data predictions to a csv."""
data = load_dataset(file_name='lending_data.csv')
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
# take a slice with no input validation issues
multiple_test_json = X_validation[99:600]
predictions = make_prediction(input_data=multiple_test_json)
predictions_df = pd.DataFrame(predictions)
# save file to lending model package, not installed package
predictions_df.to_csv(f'{config.PACKAGE_ROOT.parent}/'
f'lending_model/lending_model/datasets/{save_file}')
def test_prediction_endpoint_returns_prediction(flask_test_client):
# given
# load data from lending model package, & get validation set
data = load_dataset(file_name=model_config.TRAINING_DATA_FILE)
data = clean_target_variable(data)
_, X_validation, _, _ = split_dataset(data)
post_json = X_validation[0:1].to_json(orient='records')
# when
response = flask_test_client.post('/v1/predict/lending_rate',
json=post_json)
# then
assert response.status_code == 200
response_json = json.loads(response.data)
prediction = response_json['predictions']
response_version = response_json['version']
assert prediction[0] > 0
assert response_version == _version