none
Is it possible to deploy custom built R Models (say mlr package models) using mrsdeploy in Microsoft R Server RRS feed

  • Question

  • I have built models using mlr package in R

    mrsdeploy uses rx functions and models

    Below is a custom model

    logistic.learner <- makeLearner("classif.logreg",predict.type = "response")

    cv.logistic <- crossval(learner = logistic.learner,task = trainTask,iters = 3,stratify = TRUE,measures = acc)

    #train model
    > fmodel <- train(logistic.learner,trainTask)
    > getLearnerModel(fmodel)

    #predict on test data
    > fpmodel <- predict(fmodel, testTask)

    May i please know how to deploy this model

    Thanks

    Prem

    May i please know 

    Wednesday, April 10, 2019 2:02 AM

All replies

  • I recommend that you take a look at the detailed operationalization example listed on this page:

    https://docs.microsoft.com/en-us/machine-learning-server/operationalize/quickstart-publish-r-web-service

    It has an easy to follow example of how to deploy a model in R.

    Hope this helps.

    Thursday, April 11, 2019 8:25 PM