LAE

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dglavastorm

Hi everyone, I'm looking for a way to create an 8 week rolling dataset. Every week I import existing data then add new data to the left fields. All columns have the week ending Friday date added as a field prefix with that week's date so it looks like below/attached picture.

What I'm looking for is to remove all columns that are more than 8 weeks old based on this prefix. All dates are W/E Friday so always 7 days apart.

LS EX.jpg

Cheers,

DaveLS EXAMPLE.zip

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: ryeh

See attached graph. We use the Metadata node to first get the column names. A filter node is used to check to see if the field names contain a date. If so, a calculation is done to get the difference between today (system date) and the date in the field name. The split node gets only the fields within a certain range. In this case, I chose 4 weeks, as the dates in the field name span 8 weeks. The last step is to use the Change Metadata node.
Attachments:
wthin_n_ weeks.brg

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dglavastorm

Thanks Ryeh, will check this out.
What does the "out = TRUE" line do ?

eg.
if FieldName.isDate("CCYY-M-D")
then {
out = TRUE

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: ryeh

Ah, good catch. That was actually leftover logic that can be removed. I originally used it to keep track of which fields to emit. So even if a field did not contain a date in the name, I still want to emit it. But I took on a different approach and just used dateDiff as a check. So if a field did not contain a date in the name, I set the dateDiff to 0. That way, it always gets included in the following split node. I've attached an updated graph with the cleaned up logic.
Attachments:
wthin_n_ weeks_1.brg

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dglavastorm

Worked perfectly as always, thanks again.