maratonadev-la / desafio-2-2020-tortuga-code Goto Github PK

como puedo superar este mensaje de error
Instance creation of size small and type scikit-learn0.20 failed.

WMLClientError Traceback (most recent call last)
in
8 deployment_type='online', # No cambie este parámetro
9 deployment_format='Core ML', # No cambie este parámetro
---> 10 meta_props=model_meta # No cambie este parámetro
11 )

/opt/conda/envs/Python36/lib/python3.6/site-packages/watson_machine_learning_client/deployments.py in create(self, artifact_uid, name, description, asynchronous, deployment_type, deployment_format, meta_props, **kwargs)
520 else:
521 print_text_header_h2(u'Deployment creation failed')
--> 522 self._deployment_status_errors_handling(deployment_details, 'creation')
523 elif response.status_code == 201:
524 deployment_details = response.json()

/opt/conda/envs/Python36/lib/python3.6/site-packages/watson_machine_learning_client/deployments.py in _deployment_status_errors_handling(self, deployment_details, operation_name)
52 raise WMLClientError('Deployment ' + operation_name + ' failed. Error: ' + str(deployment_details['entity']['status_message']))
53 except WMLClientError as e:
---> 54 raise e
55 except Exception as e:
56 self._logger.debug('Deployment ' + operation_name + ' failed: ' + str(e))

/opt/conda/envs/Python36/lib/python3.6/site-packages/watson_machine_learning_client/deployments.py in _deployment_status_errors_handling(self, deployment_details, operation_name)
47 else:
48 print(error)
---> 49 raise WMLClientError('Deployment ' + operation_name + ' failed. Errors: ' + str(errors))
50 else:
51 print(deployment_details['entity']['status_message'])

WMLClientError: Deployment creation failed. Errors: [{'code': 'error_in_instance_creation', 'message': 'Instance creation of size small and type scikit-learn0.20 failed.'}]

algún consejo? llevo un buen rato y se me acabaron la ideas.

Saludos,

Tengo una duda de como considerar nuevas columnas hechas de los datos en las transformaciones para el notebook 2, ya que se hace el split de los features antes de llamar la transformación dentro del pipeline.