Create and Populate Time Dimension

Question

Create and Populate Time Dimension

patilds

SSChasing Mays

Points: 615
More actions
September 22, 2007 at 10:53 am

#374568

Comments posted to this topic are about the item Create and Populate Time Dimension

Viewing 13 posts - 1 through 12 (of 12 total)

You must be logged in to reply to this topic. Login to reply

silly_king SSC Journeyman Points: 88 More actions · Answer 1

This can be hugely useful, although for accuracy's sake i'd suggest renaming it the "Date Dimension", since it doesn't really have anything to do with Time... A Time dimension would deal with Hours/Minutes/Seconds (and that's what i was looking for).

🙂

chenthor Right there with Babe Points: 779 More actions · Answer 2

I was playign with this and noticed that the table population code misses setting the IsHoliday flag. Since the table has set that column to not null, the insert fails.

colin Robinson-345240 Mr or Mrs. 500 Points: 567 More actions · Answer 3

i would also probably drop the table rather than truncate in this instance. You would then get the same Identity value for each date each time you ran the code.

I have a question though, isnt there a performance impact of creating dim_time in this way with say a five year end date, as drop downs etc in front ends would be larger than required, and each date would create a null value in the fact table for every measure as the cube processes

Im thinking we should use code like the example in a slowly changing dimension with code in an if block

ie if Getdate() >= (select max(date) from DimTime) then

exec createDates 30(x num) days more data

an advantage to this is we would never have to revisit the time dimension unless there was a fundamental change like the Fiscal year dates changed and even this would just be a change to the sproc for future dates.

Jeff Moden SSC Guru Points: 1004290 More actions · Answer 4

colin Robinson (4/20/2008)
i would also probably drop the table rather than truncate in this instance. You would then get the same Identity value for each date each time you ran the code.
I have a question though, isnt there a performance impact of creating dim_time in this way with say a five year end date, as drop downs etc in front ends would be larger than required, and each date would create a null value in the fact table for every measure as the cube processes
Im thinking we should use code like the example in a slowly changing dimension with code in an if block
ie if Getdate() >= (select max(date) from DimTime) then
exec createDates 30(x num) days more data
an advantage to this is we would never have to revisit the time dimension unless there was a fundamental change like the Fiscal year dates changed and even this would just be a change to the sproc for future dates.

Truncation of a table results in the IDENTITY column being reset to its original seed... there is no need to drop the table to achieve this result.

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

colin Robinson-345240 Mr or Mrs. 500 Points: 567 More actions · Answer 5

Thanks for the pointer,

That would make a decent interview question. and it least a truncate means you can retain permissions etc.

khusnawi Grasshopper Points: 11 More actions · Answer 6

sorry, 1'm from indonesian, little speak english 😀

your source code, i'm excute values in populate Time Dimension Null ?

create dim time succes but value blank...

thanks....

edomiobo-795748 SSC Enthusiast Points: 141 More actions · Answer 7

Hello. I'm also a newbie. You'll have to execute the stored procedure that comes with the code by typing 'exec sp_createTimeDim' and clicking the execute button.

You'll get an error because no value has been instantiated for the isholiday column, so go to the design view and set it to null. This will, at least for now, enable you to execute the code and have a feel of the results;)

Jeff Moden SSC Guru Points: 1004290 More actions · Answer 8

colin Robinson (4/21/2008)
Thanks for the pointer,
That would make a decent interview question. and it least a truncate means you can retain permissions etc.

Thanks for the feedback... Correct... no need to rebuild any permissions that way...

Not sure it makes such a great interview question... if too many people ask it, others hear of it and it becomes one of those things people simply memorize without ever having done it. I rather like weaving it into a simple multi-statement code example and ask them questions about what the intermediate and final result set's would be. Sure, memorization could still help there, but at least I know that they know when it applies and what it actually does and when.

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

Jeff Moden SSC Guru Points: 1004290 More actions · Answer 9

colin Robinson (4/20/2008)
i would also probably drop the table rather than truncate in this instance. You would then get the same Identity value for each date each time you ran the code.
I have a question though, isnt there a performance impact of creating dim_time in this way with say a five year end date, as drop downs etc in front ends would be larger than required, and each date would create a null value in the fact table for every measure as the cube processes
Im thinking we should use code like the example in a slowly changing dimension with code in an if block
ie if Getdate() >= (select max(date) from DimTime) then
exec createDates 30(x num) days more data
an advantage to this is we would never have to revisit the time dimension unless there was a fundamental change like the Fiscal year dates changed and even this would just be a change to the sproc for future dates.

Just as a suggestion...

It seems to me that a Calendar table of this nature (time dimension, whatever), should be made to go out to 01/01/2100 and that the GUI should have criteria in it to return the correct number of dates.

Heh... as a side bar, I have to ask why anyone would use a loop to creat such a simple table... then I saw the date this was published and that pretty well explains it. 😉

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

Jeff Moden SSC Guru Points: 1004290 More actions · Answer 10

Thought I'd revisit this one... calculation for Leap Years can be a lot more simple...

ISDATE(STR(@Year,4)+'0229')

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

Paul Paiva SSCrazy Points: 2237 More actions · Answer 11

Thanks for the code - nice and clean, with great comments. I'm using it and appreciate it.

- Paul

http://paulpaivasql.blogspot.com/

suedman Grasshopper Points: 17 More actions · Answer 12

Hello,

everthing works fine after i edited the isHoliday column 🙂

Nice script!