With the latest release of Azure Machine Learning, we have updated and simplified the web service publishing workflow. With this change, users must generate a Scoring Experiment from a Training Experiment before the web service can be published. We automated the steps for creating a Scoring Experiment by adding web service input and output modules. We also automated saving the Trained Model so you no longer need to manually save a Trained Model and add it to a Scoring Experiment.
For more information, visit: Changes to the web service publishing process (Operationalization)