def test_delete_endpoint_only():
sagemaker_session = empty_sagemaker_session()
predictor = Predictor(ENDPOINT, sagemaker_session=sagemaker_session)
predictor.delete_endpoint(delete_endpoint_config=False)
sagemaker_session.delete_endpoint.assert_called_with(ENDPOINT)
sagemaker_session.delete_endpoint_config.assert_not_called()
def test_delete_endpoint_with_config():
sagemaker_session = empty_sagemaker_session()
sagemaker_session.sagemaker_client.describe_endpoint = Mock(
return_value={"EndpointConfigName": "endpoint-config"})
predictor = Predictor(ENDPOINT, sagemaker_session=sagemaker_session)
predictor.delete_endpoint()
sagemaker_session.delete_endpoint.assert_called_with(ENDPOINT)
sagemaker_session.delete_endpoint_config.assert_called_with(
"endpoint-config")
def main():
image_name = "sagemaker-sklearn-rf-regressor-local"
# Prepare data for model inference - we use the Boston housing dataset
print('Preparing data for model inference')
data = fetch_california_housing()
X_train, X_test, y_train, y_test = train_test_split(data.data,
data.target,
test_size=0.25,
random_state=42)
# we don't train a model, so we will need only the testing data
testX = pd.DataFrame(X_test, columns=data.feature_names)
# Download a pre-trained model file
print('Downloading a pre-trained model file')
s3.download_file(
'aws-ml-blog',
'artifacts/scikit_learn_bring_your_own_model/model.joblib',
'model.joblib')
# Creating a model.tar.gz file
tar = tarfile.open('model.tar.gz', 'w:gz')
tar.add('model.joblib')
tar.close()
model = Model(image_uri=image_name,
role=DUMMY_IAM_ROLE,
model_data='file://./model.tar.gz')
print('Deploying endpoint in local mode')
endpoint = model.deploy(initial_instance_count=1,
instance_type='local',
endpoint_name="my-local-endpoint")
predictor = Predictor(endpoint_name="my-local-endpoint",
sagemaker_session=sagemaker_session,
serializer=CSVSerializer(),
deserializer=CSVDeserializer())
predictions = predictor.predict(testX[data.feature_names].head(5).to_csv(
header=False, index=False))
print(f"Predictions: {predictions}")
print('About to delete the endpoint to stop paying (if in cloud mode).')
predictor.delete_endpoint(predictor.endpoint_name)
class Predictor:
def __init__(self, endpoint_name):
self.endpoint_name = endpoint_name
self.session = get_sagemaker_session()
self.attach_predictor()
def attach_predictor(self):
self.predictor = SagemakerPredictor(endpoint_name=self.endpoint_name,
sagemaker_session=self.session,
serializer=NumpySerializer(),
deserializer=NumpyDeserializer())
def predict(self, data):
return self.predictor.predict(data)
def undeploy(self):
self.predictor.delete_endpoint()
