large nominal Attributes

TellosTellos Member Posts: 1 Contributor I
edited July 2019 in Help

Hi folks,

i'm relative new with RM.

I have a data set and one of the variable is customer-ID. Their are over 1 million different customer-IDs. Is their a way to reduce the dimension so i can use them for my analysis (neural network)? I know i can cluster the ID by other variables but maybe their are other solutions. I have heard about feature hashing.



thanks for your help and your input.


best regards




  • Options
    Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,635 Unicorn

    It isn't totally clear from your post, but it seems like what you probably need to do is to use the "Set Role" operator and set the customer id attribute to type "id".  A synthetic variable such as this is not going to have any predictive value or any other value in your analysis.  Setting its role to id will automatically exclude it from consideration from most of the other operators.  Of course you can always use the "select attributes" operator as well and then bring in only the subset of variables that are relevant to your neural network.  I am not sure what you mean by "clustering" in the context of an id variable that doesn't have any inherent meaning.


    P.S.  This post appears to be in the wrong board.  I suggest you try the "Studio forum" under the product help board for similar questions in the future, it will be seen by more folks there.

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
Sign In or Register to comment.