Options
creating perfomance criteria
Contributor I
Hello everyone,
I'm a newby and i've got an issue )
I'm using rapidminer 6.
Is it possible to create my own perfomance criteria?
What I need:
1) take prediction result for data sample.
2) depending on prediction result (class or regression value) take some attribute (which is not in model) and multiply it on some predefined number if classification(regression ) is correct(withing borders).
3) calculate the sum
Thanks a lot.
I'm a newby and i've got an issue )
I'm using rapidminer 6.
Is it possible to create my own perfomance criteria?
What I need:
1) take prediction result for data sample.
2) depending on prediction result (class or regression value) take some attribute (which is not in model) and multiply it on some predefined number if classification(regression ) is correct(withing borders).
3) calculate the sum
Thanks a lot.
0
Answers
I'm not quite sure if I get what you want to do. But I'll give it a try.
1) Use "Apply Model" operator on your data - you get new columns with prediction and confidence (for nominal).
2) Connect "Generate attributes" operator, name new attribute and enter function expression, you can check for various functions and pick attributes you wish to use.
3) Connect "Aggregate" operator pick column, calculate sum
This works for 5.3, hopefully it will work in 6 as well.
Hope this helps...
cheers