Scheduling Azure ML Batch Jobs

Azure ML APIs provide a batch interface (BES) that can be used to score a batch file. To run BES jobs on a schedule, you can use [Azure WebJobs][1]. In this post, we will go through the steps of setting up an Azure ML Batch job using a WebJob. We will configure it to run on a recurring schedule. The steps are as follows: 1. Create and deploy an Azure ML web service (see [here][2]) 2. Create a .NET console app calling BES in Visual Studio using the C# sample code (see [here][3]) 3. Create a recurring Webjob Steps 1 and 2 are covered in the linked documents above. We will cover step 3 here. • After compiling and testing the console app created in step 2, find the project’s Release folder. Then zip its contents - resulting in a Release.zip file that contains the console app’s exe and reference libraries. • Log in to https://portal.azure.com with your account • Follow the instructions [here][4] to “Create an on demand WebJob in the portal” • Under Triggers, select Scheduled, and enter the CRON Expression. In this example, we are setting the job to run every 15 minutes. For more expression examples see the “Create a scheduled WebJob using a CRON expression” section of the document. ![enter image description here][5] After saving the changes, click to run the WebJob in Azure Portal. If the job does not complete successfully, click Logs to view the error message. This is a quick an easy way to set up a recurring schedule to call Azure ML Batch APIs. [1]: https://azure.microsoft.com/en-us/documentation/articles/web-sites-create-web-jobs/ [2]: https://azure.microsoft.com/en-us/documentation/articles/machine-learning-walkthrough-5-publish-web-service/ [3]: https://azure.microsoft.com/en-us/documentation/articles/machine-learning-consume-web-services/#batch-execution-service-bes [4]: https://azure.microsoft.com/en-us/documentation/articles/web-sites-create-web-jobs/ [5]: https://operationalizatio.blob.core.windows.net/publicimages/webjobs.png