Splitting based on the values in an attribute

sectyn
sectyn New Altair Community Member
edited November 5 in Community Q&A
I have a set of rows with missing values in one attribute. I would like to split the data based on the attribute with missing values in such a way that the sets will have rows upto the row which has some value in that attribute.
 
For example if the example set is:

Att1   Att2    Att3

a      test1   
b      test2   
c      test3    val3
d      test4
e      test5
f      test6
g      test7    val7
h      test8
i        test9

So after the split, I want 2 sets:

a     test1   
b     test2  
c      test3    val3

d     test4
e     test5
f      test6
g     test7    val7

Tried this in Loop Values, but not sure how to do the split.




Tagged:

Best Answer

  • BalazsBaranyRM
    BalazsBaranyRM New Altair Community Member
    Answer ✓
    Hi @jennyclub,

    this kind of row-spanning processing is quite hard in RapidMiner.

    I would try something like this:
    1. Get the the index of non-empty Att3 values. E. g. Generate ID, Filter Examples. 
    2. Loop the filtered index values (they need to be converted into nominal) and filter the other rows based ID <= current index. 
    3. Extract the current Att3 value with Extract Macro and write it into the filtered table with Generate Attributes.

    I hope this gives you an idea.

    Regards,

    Balázs

Answers

  • BalazsBaranyRM
    BalazsBaranyRM New Altair Community Member
    Answer ✓
    Hi @jennyclub,

    this kind of row-spanning processing is quite hard in RapidMiner.

    I would try something like this:
    1. Get the the index of non-empty Att3 values. E. g. Generate ID, Filter Examples. 
    2. Loop the filtered index values (they need to be converted into nominal) and filter the other rows based ID <= current index. 
    3. Extract the current Att3 value with Extract Macro and write it into the filtered table with Generate Attributes.

    I hope this gives you an idea.

    Regards,

    Balázs

  • sectyn
    sectyn New Altair Community Member
    Thank you, Balázs. This really helped.