Testing of Automodel

ssv112ssv112 Member Posts: 17 Contributor I
edited September 22 in Help
Hi, 
I have created machine learning classification models using Automodel and want to test the sample data using retrieve operator and putting it as unlabelled data in Apply operator instead of  unlabelled data coming from split validation operator. I am getting an error as shown but no error is coming through split validation unlabelled data.

As shown, with split data validation ( for testing samples) , process runs smoothly but ... it shows 


This error comes when i use random sample data by retrieve function.. as shown..




Any help is appreciated..


Thanks

Best Answer

Answers

  • varunm1varunm1 Moderator, Member Posts: 960   Unicorn
    edited September 22
    Hello @ssv112

    Can you provide the XML code of the modified process? This might be due to the modification you did. Inside the "performance for Hold out sets' there is a Log scoring times operator if you are not logging anything here due to the changes it might throw an error like this.
    lionelderkrikor
  • ssv112ssv112 Member Posts: 17 Contributor I
    Hi varun!

    attaching xml code


    Thanks for your help!

  • varunm1varunm1 Moderator, Member Posts: 960   Unicorn
    Hello @ssv112

    Sorry, I am not sure why my RM is not retrieving your process from XML. Can you please save it and export it into .rmp file and attach it here?

    Thanks
  • ssv112ssv112 Member Posts: 17 Contributor I
    Attaching .rmp file
  • varunm1varunm1 Moderator, Member Posts: 960   Unicorn
    edited September 22
    Hello @ssv112

    This looks fine for me (logging part in process), but I cannot test as I don't have data. I have a couple of questions (If you can provide data, I can test otherwise you can debug based on my points).

    1. Can you set a breakpoint after the "Log Scoring Time" inside "performance hold out sets" by right-clicking on "Log Scoring Time" operator. Also, set a breakpoint before and after this "Scoring Time as data" to see if there is an input coming into this operator.

    (OR)

    2. If you do not want the run times. You can simply disable the "Collect Runtimes" subprocess by right-clicking it and remove Enable. This won't show the runtimes but it will run the process without this error.
    lionelderkrikor
  • mschmitzmschmitz Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 2,200  RM Data Scientist
    Hi,
    to me this looks like the ordering of the operators is wrong.  Can you please check if it is the right one?

    Cheers,
    Martin

    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
    varunm1
  • ssv112ssv112 Member Posts: 17 Contributor I
    Thanks very much varun! you guys are very helpful! Looking to learn a lot from you..



    Regards,
    ssv112
    Tghadially
Sign In or Register to comment.