none
Clarifications in creating a model. RRS feed

  • General discussion

  • Questions:

    1.How can we preprocess the data when creating a model that needs to be deployed using a web service in ML server?

    ----- Is it done through specific R code and import R code in our function or we need to implement all the preprocess code in a single function.---Please suggest

    2.After creating a model and web service for the model, how can we train the model for new data?

    3. Some useful links that help us to create API for data ingestion,training,testing and how it can be done.

    Thanks in advance!!

    Regards

    Shankar


    shankar

    Friday, November 2, 2018 7:13 AM

All replies

  • Hi Shankar,

    Here are some links to example code and tutorials, that should answer some of your questions:

    https://docs.microsoft.com/en-us/machine-learning-server/deployr/deployr-application-developer-getting-started#developer-tools

    Tutorials:

    https://github.com/Microsoft/java-example-rbroker-basics

    https://github.com/Microsoft/java-example-client-data-io

    Preprocesing of the data can certainly be done in R before you call the 'publishService()' function from R.

    Take a look at the 'mrsdeploy' package:

    https://docs.microsoft.com/en-us/machine-learning-server/operationalize/how-to-deploy-web-service-publish-manage-in-r

    Hope this helps.

    Thursday, November 8, 2018 7:16 PM
  • Hi Team

    Thanks for your reply.

    Can you please let me know the possible ways of creating a web service and deploying as a model in Azure.

    The code is already developed in R .I see a lot of sample deployments for python.Please help.

    Thanks

    Shankar


    shankar

    Thursday, November 15, 2018 5:38 AM