That's what the CROSS APPLY basically does.  Test out the code I wrote to see if it accomplishes what you need.  It basically converts all the values in the comma-delimited...

Okay, here's my solution assuming you can't modify the table. Sometimes you just have to live with a bad design. If the table is really big, it might be worth...

So all you want to do is copy the table name and the number of records into a table? Why not query the  system tables? Sort of like this:

            
November 15, 2023 at 8:01 am
            
#4327862

            
            
                                                
        

Oh, owww... Normalization is your friend.  You should read Jeff Moden's article on his function DelimitedSplit8K

basically, you have to create the function in a database somewhere. (I'm in TempDB...

• This reply was modified 1 year ago by  pietlinden. Reason: "Answer the question, Claire!"

Okay, I'm working on a smaller copy (just the top like 1000 rows), and it's working except everything is imported delimited by double quotes. (Nice and fast though...) After that...

• This reply was modified 1 year ago by  pietlinden.

Can I do the whole import without using the Wizard, which creates (I think) and SSIS task? Because that crashes with Out of Memory errors?

For grins, (and because I'm used...

• This reply was modified 1 year ago by  pietlinden.

Oh, sorry. Left a lot of stuff out.  The files are getting imported into PowerBI (not in Fabric, or whatever MSFT calls it these days), and they do all the...

• This reply was modified 1 year ago by  pietlinden.

Okay, silly me... I completely missed Reza Rad's video on it, "Aggregation to speed up the performance of a Power BI report even if all IMPORTED"

You basically create an...

if all else fails, read the manual:

pandas.DataFrame.mean — pandas 2.1.1 documentation (pydata.org)

In particular, the diagnosis and procedure dimensions could be handled as a multi-valued dimension

How do you mean? Could you give an example of "a diagnosis being handled as a multi-valued...

If you're going to do lots of date-related analysis, I'd just create a Calendar table with all the columns you want to group by, and then join to that in...

Can't see your data. Without that, I'm pretty sure we can't do anything. Could you post at least some representative data? (CREATE TABLE and INSERT scripts, not pictures =) )

Never mind. I'm a knucklehead. Bring in all the tables (seems I shouldn't have to?), filter by name, expand tables, append.  "Use the force, Luke!"

• This reply was modified 1 year, 4 months ago by  pietlinden.

Why do you want to partition the table? Partitioning a table is for speeding up backup, not speeding up your queries.

(Where's the "SQL Howlers" article that Gail Shaw wrote about...

Use ROW_NUMBER() OVER (PARTITION BY<field list> ORDER BY <field>) where the columns in the PARTITION BY section are the columns you would use to determine uniqueness.

Standard way to do this...

• This reply was modified 1 year, 4 months ago by  pietlinden.