Charge Id – Deploying a ML.Net Model to Azure

In the previous post we built a machine learning model using ML.Net, in this post we will deploy the model to an Azure app and allow it to be used via a HTTP API

Using the output model in zip format ‘vita-model-1.zip’ we can include this in our web application as an embedded resource or simply include the file for deployment.

To use the file from a HTTP endpoint:

  1. Include the zip file in your deployment – embedded resource/content/read from blob storage etc..
  2. Initialise the model as a singleton during application start up by using a file path or stream
  3. Call the model using the function PredictionModel.Predict(‘my data from which to predict’)

Sample below that logs to Logz.io

 [Produces("application/json")]
    [Route("[controller]")]
    public class PredictionController : Controller
    {
        private readonly IPredict _predictor;

        public PredictionController(IPredict predictor)
        {
            _predictor = predictor;
        }

        [HttpPost("predict/")]
        [SwaggerResponse(HttpStatusCode.OK, typeof(string))]
        public async Task<IActionResult> Search(PredictionRequest request)
        {
            Guard.AgainstNull(request);
            var requestId = Guid.NewGuid();
            using (LogContext.PushProperty("request", request.ToJson()))
            using (LogContext.PushProperty("requestId", requestId))
            {
                try
                {
                    var result = await _predictor.PredictAsync(request);
                    return Ok(result);
                }
                catch (Exception e)
                {
                    Console.WriteLine(e);
                    Log.Warning(e, "PredictionController error {request}", request.ToJson());
                    return NoContent();
                }
            }
        }
    }

Hosting our endpoint with Swagger on Azure allows us to test the inputs and see the results below:

Timage

Conclusion

Here we hosted our model in Azure using an App Service and managed to test it via Swagger.

Hoping to make this a Function App when this issue is resolved –> https://github.com/dotnet/machinelearning/issues/569

 

 

 

Posts in this series

Charge Id – scratching the tech itch [ part 1 ]
Charge Id – lean canvas [ part 2 ]
Charge Id – solution overview [ part 3 ]
Charge Id – analysing the data [ part 4 ]
Charge Id – the prediction model [ part 5 ]
Charge Id – deploying a ML.Net Model to Azure [ part 6 ]

 

Code

https://github.com/chrismckelt/vita