Hi dear,

If I have multiple fields like:

CustID  Items_A_Bought        Items_B_Bought ... Items_Z_Bought

John    1                                 7                          0

Jack    0                                0                           1

Jess    2                                2                           2

 And I want to calculate how many types of items they bought:

John  2 (A and B)

Jack 1 (C)

Jess 3 (A,B, C)

 The way I can think of is to use Filler node to fill where @FIELD>0 with 1, then sum_n(@fields_between(Item_A_Bought,Item_Z_Bought). Is there any other way to do this?

Thanks,

Your method seems fine.

Most of my analysis is on a database, so I would solve this data analysis a little differently.  See the attached zipped stream+data, it will pushback the processing as SQL when you are connected to a database (I'm not sure if the functions  "sum_n(@fields_between"  would work as SQL).

My alternative hardcodes the field names and could be tedious if you have a lot of fields, but a script can be used to loop through all the fields and build the expression dynamically if you have hundreds of fields.

I also used an aggregate in case you have the same customer on many rows.  You can remove the aggregate node if you want the customer transactions to be separate.

Cheers

Tim

 

You can try this:

count_greater_than(0, @FIELDS_BETWEEN(Item_A_Bought,Item_Z_Bought))

So.. you didn't need to use Filler node anymore.

Umar

SPSS Indonesia