Comments on: PowerPivot time intelligent functions revisited: why use ALL() and how to work around it

By: Kasper de Jonge

Kasper de Jonge — Sat, 11 Sep 2010 06:26:59 +0000

Hi @Bobby Henningsen , I think we all never use a datetime as the key. There are several ways to work around this, you can create one using =date() if you have year, month, day available in you fact table.

By: Bobby Henningsen

Bobby Henningsen — Sat, 11 Sep 2010 06:01:32 +0000

Spot on. I of course missed the import thing about relationsship with a date column. Which of course makes things a little bit harder since i never use date as key in my time dimension Thanks..

By: Kasper de Jonge

Kasper de Jonge — Fri, 10 Sep 2010 20:35:15 +0000

Hi @Bobby Henningsen ,

The two most important things to do when you want to work with time intel functions is
- Create a separate related time table (dimension)
- Use a date column to relate to the separate related time table

Check out this blog post as well:
http://www.powerpivotblog.nl/powerpivot-time-intelligent-functions-golden-rules

I think this will help you get the time intel functions to work as expected, the all function shouldn’t be necessary anymore.

By: Bobby Henningsen

Bobby Henningsen — Fri, 10 Sep 2010 15:25:15 +0000

Actually it will work with the “ALLExcept” but it’s kind off “Not that intuitive” But I am kind of worried that the “Books online” isn’t correct. Can you confirm that..

(ps I’m danish

By: Bobby Henningsen

Bobby Henningsen — Fri, 10 Sep 2010 13:45:25 +0000

I’ve starting playing around with the Parallelperiod in DAX. I’ve been using MDX for some years now so I am used to the Paralleperiod function but I face a lot of strange behaviours comparing to what I am used to First of all I am not able to use it unless i do the “ALL” thing. Books online suggests the following : =CALCULATE(SUM(InternetSales_USD[SalesAmount_USD]), PARALLELPERIOD(DateTime[DateKey],-1,year)) but that doesn’t work at all. It needs the ALL. But this will ignore the filters. Using ALLEXCEPT will make this work. But how can I make it work when I want to use Quarters in my rows as well ? Using the ALLExcept ? Great Blog BTW

By: SQLBI - Marco Russo : Hide the Date column whenever you define a Dates table in PowerPivot

Tue, 29 Jun 2010 15:13:59 +0000

[...] are already several posts in the blogosphere about the need of using a separate Dates table in PowerPivot to make almost any [...]

By: Kasper de Jonge PowerPivot and MS BI Blog » PowerPivot Time intelligent functions golden rules

Thu, 13 May 2010 07:11:43 +0000

[...] PowerPivot time intelligent functions revisited: why use ALL() and how to work around it Create a separate related time table for time intelligent functions in PowerPivot Screencast: Introduction to time intelligent functions in PowerPivot [...]

By: Kasper de Jonge

Kasper de Jonge — Wed, 14 Apr 2010 07:32:02 +0000

@johncon
Hi John,

I have made a new blog post on creating a separate related time table and how to use this in RTM and CTP:
http://www.powerpivotblog.nl/create-a-separate-related-time-table-for-time-intelligent-functions-in-powerpivot

You were almost there as i can see from your questions, I hope my new post has answered your questions.

For Q3 the all() indeed isn’t needed when you use the date field from your related time table as the date column to be calculated upon in your time intelligent function (as you can see in the example of my new blog post).

I hope this answers your questions, if not don’t hesitate to ask!

Kasper
(ps i’m Dutch )

By: Kasper de Jonge PowerPivot & MS BI Blog » Create a separate related time table for time intelligent functions in PowerPivot

Wed, 14 Apr 2010 07:01:44 +0000

[...] This will make sure you don’t need all kind of tricks to make sure you have date, check out this blog post to see why we need a related time table in more detail. This blog post will describe how to create [...]

By: johncon

johncon — Tue, 13 Apr 2010 19:26:35 +0000

Uggg. I swear a time-lord would have fits over these things. I am not sure if I understand the work around properly (or if I can even get it to work). My sales (aka facttable) has the date field in it and I have multiple entries for the same date. It does have a unique – record ID field.
I made a copy of the Sales table, stripped out most of the extra columns and left the RecordID and Sales Date field. I think linked Sales with this called DimSales on the RecordID field.

Q1. Despite the new date table, the ALL() still removed all filters from preset pivottables. Do I need to have it linked by the Date fields in order for this to work?
Q2. As I can not have distinct dates on my sales table, will putting distinct dates on my DimSales table omit sales entries as I have many sales per date?
Q3. You state that with the RTM and future roll-outs, the necessity of the ALL() for time measures won’t be needed, but you state that a separate time table is still required. (a) Why? (b) How do you incorporate the time table in the measures if the ALL command won’t be necessary. (c) If you still need the time table, then what’s the point of removing ALL() as a necessary parameter?

Sorry – you have answers (and a lovely Danish BI title now I believe – congrats!) that I need. Thanks.