Comments on: Year-to-Date Query Improvement Through Transformation Tables http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/ Oracle Data Warehouse Design and Architecture Fri, 04 Jul 2008 18:52:44 +0000 http://wordpress.org/?v=MU By: David Aldridge http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51044 David Aldridge Tue, 10 Jun 2008 15:33:37 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51044 The issue here is that although it is expedient to code a non-equijoin to implement a metric such as a year-to-date sales sum, it is generally better to create a table to implement the requirement as equi-joins because that allows the optimizer to implement the query using the extremely low cost hash join. As for the name, I'm not particularly attached to "transformation" ... a rose by any other name would smell as sweet, right? I seem to recall that the last time I used Microstrategy they refered to these types of metrics as "transcendental". The issue here is that although it is expedient to code a non-equijoin to implement a metric such as a year-to-date sales sum, it is generally better to create a table to implement the requirement as equi-joins because that allows the optimizer to implement the query using the extremely low cost hash join.

As for the name, I’m not particularly attached to “transformation” … a rose by any other name would smell as sweet, right? I seem to recall that the last time I used Microstrategy they refered to these types of metrics as “transcendental”.

]]> By: Lodewijk Wiggers http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51043 Lodewijk Wiggers Tue, 10 Jun 2008 14:26:15 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51043 Transformation tables?? I hope i am not misunderstanding this, but you are just refering to having smart time/date dimensions which make these kinds of conditional summing time/date wise possible date year period month first_day_of_year first_day_of_month first_day_of_week previous_workday 20080609 2008 200806 06 20080101 20080301 20080609 20080606 20080610 2008 200806 06 20080101 20080301 20080609 20080609 20080611 2008 200806 06 20080101 20080301 20080609 20080610 ... This can be extended to all the usefull conditional summing you might need, as in for example the normally hard to solve business day summing: give me the total of the last 5 business days... Having pre-generated time/date dimensions like this makes conditional summing so much easier. Just don't understand why people call them 'transformation' tables, that just doesn't make sense. What is there to 'transform'? Transformation tables?? I hope i am not misunderstanding this, but you are just refering to having smart time/date dimensions which make these kinds of conditional summing time/date wise possible

date year period month first_day_of_year first_day_of_month first_day_of_week previous_workday
20080609 2008 200806 06 20080101 20080301 20080609 20080606
20080610 2008 200806 06 20080101 20080301 20080609 20080609
20080611 2008 200806 06 20080101 20080301 20080609 20080610
…

This can be extended to all the usefull conditional summing you might need, as in for example the normally hard to solve business day summing: give me the total of the last 5 business days…

Having pre-generated time/date dimensions like this makes conditional summing so much easier.

Just don’t understand why people call them ‘transformation’ tables, that just doesn’t make sense. What is there to ‘transform’?

]]> By: akhil http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51036 akhil Tue, 03 Jun 2008 06:12:22 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-51036 hai all i want pervious quarter start date and end date hai all i want pervious quarter start date and end date

]]> By: Rittman Mead Consulting » Blog Archive » Year-to-date alternatives http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-50815 Rittman Mead Consulting » Blog Archive » Year-to-date alternatives Mon, 17 Mar 2008 14:25:24 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-50815 [...] Aldridge challenged me to write about functions that return lists of dates. So here [...] [...] Aldridge challenged me to write about functions that return lists of dates. So here [...]

]]> By: Joe http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-35154 Joe Tue, 17 Jul 2007 20:57:17 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-35154 I can't get to the complete test script, and the result of an execution on Oracle 10.2.0.2.0. Please correct link. thanks. I can’t get to the complete test script, and the result of an execution on Oracle 10.2.0.2.0. Please correct link. thanks.

]]> By: Time series problems « Pete-s random notes http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-30389 Time series problems « Pete-s random notes Thu, 14 Jun 2007 19:19:15 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-30389 [...] series problems A while back David Aldridge and I both blogged about the joys of time series analysis on sparse data; that is, the sort of data [...] [...] series problems A while back David Aldridge and I both blogged about the joys of time series analysis on sparse data; that is, the sort of data [...]

]]> By: Stephen http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-12127 Stephen Wed, 14 Mar 2007 09:29:07 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-12127 it is clear to me too!!! thanks it is clear to me too!!!

thanks

]]> By: Rogier Werschkull http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2729 Rogier Werschkull Wed, 11 Oct 2006 15:31:55 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2729 Thank you David, it is clear to me now! Thank you David, it is clear to me now!

]]> By: David Aldridge http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2727 David Aldridge Wed, 11 Oct 2006 14:34:52 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2727 Rogier, The month always has to be included in the select clause, and if you have any other higher dimensional columns in the YTD table they would be purely for predicating -- for example, "give me the year-to-month totals for all months in quarter 2006 Q3". If you wanted to apply the same method to year-to-quarter queries you'd use a different table .... 2006Q1 2006Q1 2006Q2 2006Q1 2006Q2 2006Q2 2006Q3 2006Q1 2006Q3 2006Q2 2006Q3 2006Q3 2006Q4 2006Q1 2006Q4 2006Q2 2006Q4 2006Q3 2006Q4 2006Q4 Rogier,

The month always has to be included in the select clause, and if you have any other higher dimensional columns in the YTD table they would be purely for predicating — for example, “give me the year-to-month totals for all months in quarter 2006 Q3″.

If you wanted to apply the same method to year-to-quarter queries you’d use a different table ….

2006Q1 2006Q1
2006Q2 2006Q1
2006Q2 2006Q2
2006Q3 2006Q1
2006Q3 2006Q2
2006Q3 2006Q3
2006Q4 2006Q1
2006Q4 2006Q2
2006Q4 2006Q3
2006Q4 2006Q4

]]> By: Rogier Werschkull http://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2719 Rogier Werschkull Wed, 11 Oct 2006 09:34:42 +0000 https://oraclesponge.wordpress.com/2006/08/15/year-to-date-query-improvement-through-transformation-tables/#comment-2719 I don't quite get it, take the example below for data that has a yearmonth grain. if i would like the yearmonth totals for 2006Q1 and 2006Q2 i would get wrong results because I would get duplicates of the yearmonth in ytd column that will be joined to the facts... regards, Rogier Werschkull ------------ <b>yearmonth in ytd yearmonth quarter</b> 200601 200601 2006Q1 200601 200602 2006Q1 200602 200602 2006Q1 200601 200603 2006Q1 200602 200603 2006Q1 200603 200603 2006Q1 200601 200604 2006Q2 200602 200604 2006Q2 200603 200604 2006Q2 200604 200604 2006Q2 200601 200605 2006Q2 200602 200605 2006Q2 200603 200605 2006Q2 200604 200605 2006Q2 200605 200605 2006Q2 200601 200606 2006Q2 200602 200606 2006Q2 200603 200606 2006Q2 200604 200606 2006Q2 200605 200606 2006Q2 200606 200606 2006Q2 I don’t quite get it, take the example below for data that has a yearmonth grain. if i would like the yearmonth totals for 2006Q1 and 2006Q2 i would get wrong results because I would get duplicates of the yearmonth in ytd column that will be joined to the facts…

regards,
Rogier Werschkull
————
yearmonth in ytd yearmonth quarter
200601 200601 2006Q1
200601 200602 2006Q1
200602 200602 2006Q1
200601 200603 2006Q1
200602 200603 2006Q1
200603 200603 2006Q1
200601 200604 2006Q2
200602 200604 2006Q2
200603 200604 2006Q2
200604 200604 2006Q2
200601 200605 2006Q2
200602 200605 2006Q2
200603 200605 2006Q2
200604 200605 2006Q2
200605 200605 2006Q2
200601 200606 2006Q2
200602 200606 2006Q2
200603 200606 2006Q2
200604 200606 2006Q2
200605 200606 2006Q2
200606 200606 2006Q2

]]>