Options

Options - Pivot

choose_usernamechoose_username Member Posts: 33 Contributor II
Hello,

i have a short question about the Operator Pivot. What does that the Option

Consider Weights do ? What weights are these? in the next option i can configure
how that weights should be summed. ...

if someone could explain that to me i would be glad.

greetings

choose_username

btw i have already read the discription ....

Answers

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

    I must admit I have never used this option myself and I am not 100% sure but anyway... Quote from the documentation: "If the source example set contains example weights, these weights may be aggregated in each group to maintain the weightings among groups."

    As far as I understand this (and the source code at a short glance), this basically means that example weights (encoded by an attribute with the role "weight") will be aggregated in each group with the specified function and the resulting weight is used for the new examples.

    Cheers,
    Ingo

  • Options
    choose_usernamechoose_username Member Posts: 33 Contributor II
    @ I must admit I have never used this option myself and I am not 100% sure but anyway

    this makes u just more likeable  :D
    ______________________________

    if i understand this correctly does it do the following?!?!:

    if i have an example set  A | B | C  with the weights  1 | 2 | 1

    A | B | C
    1  | 0 | 0
    0  | 1 | 1
    1  | 1 | 1

    then the aggregation (Sum for example) wont be 


    A | B | C
    2  | 2 | 2

    it will be

    A | B | C
    2  | 4 | 2
    if this option is activated?

    Could this be correct?  i have to think to test this...

    greetings

    choose_username


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

    Could this be correct?  i have to think to test this...
    Hmm, I don't think so.

    The question is more: if you have a special attribute "weight" giving each example a weight, how would you handle this? Just skip it? I have created a demo process and have uploaded it with the Community Extension under the name

    "Pivoting with Consideration of Example Weights"

    This process shows how the Pivoting operator can also consider example weights encoded by a special attribute with the role 'weight'. The first operators only create a demo data set for this purpose, the actual pivoting is performed by the last operator only. I recommend to place a breakpoint before the Pivoting and check the data (structure) in order to see how the Pivoting handles the weights.

    You could also aggregate the weights, use the group as id and join the aggregated weights to a pivoted set (with the group as id, too). The parameter just makes things a bit more convenient here.

    Cheers,
    Ingo
  • Options
    choose_usernamechoose_username Member Posts: 33 Contributor II
    yesterday i recognized that my approach is complete nonsense. i downloaded ur workflow and analysed it.

    it seems to me that this option (consider weights) just have an effect only when an Attribute is available with the role Weight. and the functions are just the way the Attributes is treated...

    I hope that i have understood this.


    greetings

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

    it seems to me that this option (consider weights) just have an effect only when an Attribute is available with the role Weight.
    that's right.

    Cheers,
    Ingo
  • Options
    choose_usernamechoose_username Member Posts: 33 Contributor II
    ok then thanks alot for the infomation

    greetings

    choose_username
Sign In or Register to comment.