We would like to analyze student test score data as well as strand score
data.
Sample data may look like this:

Math Test Raw Score: 1590 (Test)
Shapes: 250 (Strand)
Spaces: 185 (Strand)
Measurements: 284 (Strand)

Problem: There is a random number of strands between zero and ten. We are
wondering how to model this process. Is there a way to analyze this
information in one fact table so that we can see score and strand
information?

sample_fact_table (horizontal)
----------------------------------
raw_score | strand_1 | strand_2 | strand_[n]
1590 | 250 | 185 | 284

sample_fact_table (vertical)
----------------------------------
raw_score
1590
250
185
285

I don't know what "strand" is, but in the abstract you might consider a
normalized design where rather than having columns in the fact table for
Strand0 - Strand10, instead have a StrandNo column and use it as the basis
for a Strand dimension.

public @ the domain below
www.tomchester.net

"Rick" <bob (AT) bob (DOT) net> wrote

Can you expand on this? I can't visualize it. I was thinking of a
parent-child dimension for the unbalanced nature. Does that make sense?


"Tom Chester" <nospam_public (AT) tomchester (DOT) net> wrote

It's hard to give sound advice with sketchy knowledge. That said, instead of
your fact table containing columns Strand1, Strand2, etc, it would contain
one column - StrandID -- containing a value of 0-10 (or whatever). This
column would be the basis for a Strand dimension. This removes "repeating
groups" from the fact table design, and does not require a re-design if/when
Strand11 emerges.

Again, I don't know what strand is, the suggestion (to consider) is in the
abstract.

public @ the domain below
www.tomchester.net

"Rick" <bob (AT) bob (DOT) net> wrote