RM hangs at the conclusion of AutoModel

pblack476pblack476 MemberPosts:83Maven
edited November 2019 inHelp
I have been trying to run AM the whole day and when it ends it hangs and I cannot click on anything. Earlier today it was fine.

I can still save a model that finished calculating while AM is calculating the next model. But I cannot save results or anything else because the program freezes at the end.
Tagged:

Best Answer

  • pblack476pblack476 MemberPosts:83Maven
    Solution Accepted
    I have no clue as to why but on another DataSet I was able to run everything normally. I will consider my issue resolved for all intents and purposes as that specific dataset is not relevant anymore for me.
    sgenzer

Answers

  • lionelderkrikorlionelderkrikor Moderator, RapidMiner Certified Analyst, MemberPosts:1,195Unicorn
    Hi@pblack476,

    In order we try to reproduce what you observe, can you share your data and precise what option(s) you chose in AM ?
    Which release of RM are you using ?

    Regards,

    Lionel
  • sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM ModeratorPosts:2,959Community Manager
    嗨@pblack476你的rapidminer的副本-studio.log file would also be helpful.

    Scott
  • pblack476pblack476 MemberPosts:83Maven
    Here it is. The data and the log file. It hangs when I try using a Random Forest with any Automatic Feature Generation turned on. A Simple run of Naive Bayes finished normally. Other models also seem to cause hanging but you can test over there.

  • sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM ModeratorPosts:2,959Community Manager
    edited November 2019
    thank you@pblack476I am running the data thru AutoModel now. Note that Automatic Feature Generation is VERY resource-intensive...it is not unusual for it to take a while.

    Thx also for your log file - nothing unusual there.

    Scott

  • sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM ModeratorPosts:2,959Community Manager
    so just FYI it's still churning away with two models so far (Naive Bayes and GLM) with total elapsed time of 13+ minutes. For benchmarking purposes I'm running this on a MacBook Pro with an Intel 2.3GHz 8-core i9 processor & 32GB RAM. As you can see it's a monster with Automatic Feature Generation (hence why you get the "it will take a long time" warning")


    lionelderkrikor
  • sgenzersgenzer Administrator, Moderator, Employee, RapidMiner Certified Analyst, Community Manager, Member, University Professor, PM ModeratorPosts:2,959Community Manager
    so update: the Auto Model process completed including Random Forest. It took a while but again this is expected. FWIW from this sample I would not bother with Random Forest anyway - Logistic Regression is the clear winner.


    lionelderkrikor IngoRM
  • pblack476pblack476 MemberPosts:83Maven
    @sgenzerthank you.

    I usually just limit AFE to a few minutes to save time for the sake of comparing models and then just choose a few to run on longer periods to see if i can squeeze some performance out of it. But taking a long time is not the issue, I realize it does but the process ends because I can hear the fans on my computer stopping and that means the calculations are done. When that happens, I cannot click on anything on the program.
Sign InorRegisterto comment.