API_Request In-database Function Usage | Azure Machine Learning | Vantage - API_Request In-database Function Usage - Teradata Vantage

Usage Considerations

Usage Example

1. Set argument names in Azure CLI.

   export GROUP_NAME=<RG_Name> 
   export LOCATION=<Azure_Region_Name> 
   export WORKSPACE=<WorkSpace_Name> 
   export SUB=<Subscription_id> 
   export ENDPOINT_NAME=sample-endpoint 
   export AZ_ML_EXAMPLE_DIR=<Download azure ml example from https://github.com/Azure/azureml-examples and set it as a root directory> 
   

2. Set Azure account and Azure Machine Learning workspace in Azure CLI.

   az login 
   az account set --subscription $SUB 
   az group create --name $GROUP_NAME --location $LOCATION 
   az ml workspace create -n $WORKSPACE -g $GROUP_NAME 
   

   cd $AZ_ML_EXAMPLE_DIR/cli/endpoints/online/managed/sample 
   

3. Create endpoint on Azure Machine Learning in Azure CLI.
   • Create the endpoint.

     az ml online-endpoint create --verbose --name $ENDPOINT_NAME -f endpoint.yml 

   • Create a deployment named 'blue' under the endpoint.

     az ml online-deployment create --verbose --name blue --endpoint $ENDPOINT_NAME -f blue-deployment.yml --all-traffic 
     

   • Get the key used to authenticate the endpoint.

     az ml online-endpoint get-credentials --verbose -n $ENDPOINT_NAME -o tsv --query primaryKey 
     

   • Check the status of the endpoint and the deployment.

     az ml online-endpoint show --verbose -n $ENDPOINT_NAME -o tsv --query scoring_uri

   Once the endpoint is created, you should save the primaryKey and endpoint name.
4. Use API_Request query from Vantage to score data.
   • ENDPOINT can take a Azure Machine Learning endpoint for scoring.

     SELECT rec_id, output as model_score 
     FROM tapidb.API_Request 
     ( 
       ON (SELECT 751493.0 as "rec_id", 1,2,3,4,5,6,7,8,9,10) 
       USING AUTHORIZATION('{"Key" : "<primaryKey>",
                             "Region" : "<Azure-region-Name>"}')     
       API_TYPE('azure-ml') 
       ENDPOINT('sample-endpoint') 
       CONTENT_TYPE('json') 
       KEY_START_INDEX('1') 
       CONTENT_FORMAT('{"data" : [["%row"]]}') 
     ) as "DT" 
     ORDER BY 1,2
     ; 

     API_Request query result:

     *** Query completed. 
     One row found. 
     2 columns returned. 
     *** Total elapsed time was 3 seconds. 
     rec_id model_score --------- ------------------------------------------------------------
     751493.0 11055.977245525677 +---------+---------+---------+---------+---------+---------+

   • ENDPOINT can also take scoring URL instead of endpoint name.

     select rec_id, output as model_score 
     FROM tapidb.API_Request 
     ( 
       ON (SELECT 751493.0 as "rec_id", 1,2,3,4,5,6,7,8,9,10) 
       USING AUTHORIZATION('{"Key" : "<your_auth_key>", 
                             "Region" : "westus2"}') 
       API_TYPE('azure-ml') 
       ENDPOINT('http://f160edb9-2201-4da6-861c-78de995ab519.westus2.azurecontainer.io/score') 
       CONTENT_TYPE('json') 
       KEY_START_INDEX('1') 
       CONTENT_FORMAT('{"data" : [["%row"]]}') 
     ) as "DT" 
     ORDER BY 1,2
     ; 

     API_Request query result:

     *** Query completed. 
     One row found. 
     2 columns returned. 
     *** Total elapsed time was 3 seconds. 
     rec_id model_score --------- ---------------------------------------------------------------------
     751493.0 6141.267272547496 +---------+---------+---------+---------+---------+---------+---------+