tutorial-run-ml-model-on-sql-edge.md

title	description	author	ms.author	ms.reviewer	ms.date	ms.service	ms.topic
Deploy ML model on Azure SQL Edge using ONNX	In part three of this three-part Azure SQL Edge tutorial for predicting iron ore impurities, you'll run the ONNX machine learning models on SQL Edge.	kendalvandyke	kendalv	randolphwest	09/21/2024	azure-sql-edge	tutorial

Deploy ML model on Azure SQL Edge using ONNX

[!INCLUDE retirement-notice]

Note

Azure SQL Edge no longer supports the ARM64 platform.

In part three of this three-part tutorial for predicting iron ore impurities in Azure SQL Edge, you'll:

1. Use Azure Data Studio to connect to SQL Database in the Azure SQL Edge instance.
2. Predict iron ore impurities with ONNX in Azure SQL Edge.

Key components

1. The solution uses a default 500 milliseconds between each message sent to the Edge Hub. This can be changed in the Program.cs file

   TimeSpanmessageDelay=configuration.GetValue("MessageDelay",TimeSpan.FromMilliseconds(500));

2. The solution generated a message, with the following attributes. Add or remove the attributes as per requirements.

   { timestampcur_Iron_Feedcur_Silica_Feedcur_Starch_Flowcur_Amina_Flowcur_Ore_Pulp_pHcur_Flotation_Column_01_Air_Flowcur_Flotation_Column_02_Air_Flowcur_Flotation_Column_03_Air_Flowcur_Flotation_Column_04_Air_Flowcur_Flotation_Column_01_Levelcur_Flotation_Column_02_Levelcur_Flotation_Column_03_Levelcur_Flotation_Column_04_Levelcur_Iron_Concentrate }

Connect to the SQL Database in the Azure SQL Edge instance to train, deploy, and test the ML model

1. Open Azure Data Studio.

2. In the Welcome tab, start a new connection with the following details:

   Field	Value
   Connection type	Microsoft SQL Server
   Server	Public IP address mentioned in the VM that was created for this demo
   Username	sa
   Password	The strong password that was used while creating the Azure SQL Edge instance
   Database	Default
   Server group	Default
   Name (optional)	Provide an optional name

3. Select Connect.

4. In the File section, open /DeploymentScripts/MiningProcess_ONNX.jpynb from the folder in which you cloned the project files on your machine.

5. Set the kernel to Python 3.

Related content

• For more information on using ONNX models in Azure SQL Edge, see Machine learning and AI with ONNX in SQL Edge.