A static model

SquirrelXSquirrelX Member Posts: 9 Contributor II
edited November 2018 in Help

I'm trying to compare some of my normally trained models to a predefined, static one.
I have a x-validation box for one of the proper models that outputs all sorts of performance values I need (and that's what I want indeed). I created another x-validation box for the static model so the values outputted would be comparable. However, what I couldn't figure out how to do is to tell the model what its output should be. Basically, I have an attribute which is what I want as output for each record, and another attribute with the correct output, and I just want to see what my static outputs' performance is with exactly the same folding as for the proper models.

Could anyone please help me out with some suggestions?

(btw, rapidminer seems to manage memory pretty badly, if any other java application is open next to it everything will crash, so I either use my browser and check emails or use rapidminer, which is quite a pitty. [end of complaint])


  • Options
    ZiggiZiggi Member Posts: 8 Contributor II
    >> btw, rapidminer seems to manage memory pretty badly

    Yeap, please - have a look on my post: http://rapid-i.com/rapidforum/index.php/topic,3870.0.html

    IMHO - unless this will be fixed with the next releases, it would become the serious disadvantage of the package compared to competitive solutions. What does it mean that I can build the model easily if the application becomes unstable, interactivity is ruined and entire system is affected with massive memory consumption?

  • Options
    IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder
    Hi Ziggi,

    please have a look at my comments in your original thread and please note that you are free to adapt the memory consumption strategy to your requirement (which you would know if you have read the proper documentation  :P )

Sign In or Register to comment.