When you conduct multple experiments, you will find the most appropreate model for your buiness. And after the model selection, you need to make the score model as a service to use for your business. In this lab, we'll create a web service to expose your score model.
- Deploy a Web Service
Open a new browser or tap and go to Azure ML Studio
Based on the result, click on a model you want to use for operation, and then click 'Predictive Web Service [Recommended]' to be ready to operationalize your model
When Predictive Experiment is ready, click on 'Run' to make model ready
When running is done your screen will look like following
Delete connection between 'Import Data' module and 'Score Model' module
Add 'Select Columns in Dataset' module to canvas and click 'Lanch column selector' to filter a column
Click 'WITH RULES' and choose 'ALL COLUMNS', 'Exclude' and type 'churn'
'Run' the experiment
Click 'Deploy Web Service [Classic]' to deploy a service
Note: Make sure your exepriment ran correctly and it's status shown as Finished running on the top right conner
When deployment is done, you can download sample excel file to test your web service
To test your service, download one of excel file to your local machine
Open the downloaded excel file and click on 'enable' button to run excel add-in
Select your machine leaning model, it'll look like following
Next > 05. Run Azure ML in Batch