Question

I have a row of data as follows:

            header1      header2      header3      header4      header5
row key     datavalue1   datavalue2   datavalue3   datavalue4   datavalue5....

so basically, I have a denormalized data set where the datavalues may or may not be empty on a row-by-row basis. I need to normalize them.

ie

12345678    NULL         10           3            NULL         14

would become:

12345678   header2   10
12345678   header3   3
12345678   header5   14

I could do this by using a paste special transform, but I have thousands of rows and I’d need to make sure that I get the right row key for each. furthermore, each row has a bunch of descriptives associated with it that I need copied over with each datavalue.

What is the easiest way to convert each row of columns such that I have multiple rows of a single column with all non-empty datavalues plus the associated datavalue reference? I need to be able to pivot the dataset.

Answer 1

If you have five “header” columns, enter these formulas

H1: =OFFSET($A$1,INT((ROW()-1)/5)+1,0)
I1: =OFFSET($A$1,0,IF(MOD(ROW(),5)=0,5,MOD(ROW(),5)))
J1: =INDEX($A$1:$F$9,MATCH(H1,$A$1:$A$9,FALSE),MATCH(I1,$A$1:$F$1,FALSE))

Copy H1:J?? and paste special values over the top. Sort on column J and delete anything that’s a zero. If you have legitmate zeros in the data, then you first need to replace blank cells with some unique string that you can then delete later.

If you have more columns, then replace the ‘5’ in all the above formulas with whatever number you have.