none
Revolution Analytics R - RxExec doesn't work on Multiple Cores when running Survival Random Forest model. RRS feed

  • Question

  • I have a machine learning Random Forest Survival Analysis model and trying to run it on Multicore Cluster using Revolution Analytics  R Enterprise 7.1.0 version.

    I have used below code

    library(RevoScaleR)
    rxSetComputeContext(RxLocalParallel())
    rxOptions(numCoresToUse = 5)

    And then running the RxExec to run the random forest function using randomForestSRC package.

    I could only see CPU utilization as 10% which is by default 1 core. 

    Am I missing anything here? Does the randomForestSRC package need to be a parallel one ?


    Friday, July 22, 2016 4:10 PM