I am trying to get my head around how to design the SSAS tabular model for a scenario where I have an attribute on a dimension that will modify an otherwise straight-forward measure on a related fact table.
The fact table is currently at 155 million rows. The dimension has 6200 rows. The measure we need sums a unit volume fact, then multiplies that sum by a factor that belongs to the dimension. The context used for this measure has some constraints, then: the aggregation can never include multiple rows from the related dimension. It doesn't make any business sense to do so.
I started with this, which I know is wrong: =SUM([Unit Volume])*Products[Foo Factor], and of course the designer complains because the related attribute has many values without a specific context.
My work-around for now is to create a calculated column that brings the factor into the fact table: =RELATED(Products[Foo Factor]), and then I can create a measure that looks like this: =SUM([Unit Volume])*AVERAGE([Foo Factor]).
I don't like the idea of putting calculated columns on that fact table. It's huge and it only going to get bigger. I also don't like that there doesn't appear to be a way to put constraints on the context for a given measure so I can ensure the we don't roll up volume across multiple products. Or is there?
It seems to me that ours is not an unusual scenario, but I'm too new to DAX to know how to model it correctly. If I were in my familiar SQL world, I would use a windowing function like this:
select sum([Unit Volume]) over (partition by <context goes here>) as [Unit Volume Total]
Can I create something like this in a measure with DAX?
UPDATE
It seems like this should work, but the designer still barks:
=CALCULATE(SUM('My Facts'[Unit Volume])*Products[Foo Factor],FILTER(Products,Products[Product Code]="0"))
My thought was that the [Product Code]="0" filter would be replaced by the current context and since [Product Code] is defined as the key on the related table, SSAS would know my intent. Alas, it doesn't.
I think I may have it. Seems like I may not be able to get away from using an AVERAGE aggregation on the foo factor, but the HASONEVALUE function does enforce the constraint I want.
This should work:
=
IF (
HASONEVALUE ( Products[Product Code] ),
SUM ( 'My Facts'[Unit Volume] ) * AVERAGE ( Products[Foo Factor] ),
SUM ( [Unit Volume] )
)
Please try this:
=CALCULATE((SUM(My Facts[Unit Volume])*Products[Foo Factor]),Products[Product Code]=0)
Also please confirm Product Code format.
Related
All of the measure that I want to cumulative has the same formula. So, is there any way to use the thing like function or any thing in calculate measure to resolve this issue?
There are two ways to achieve your aim:
1- the first solution is based on using the business intelligence wizard to add time intelligence to your solution.
The time intelligence enhancement is a cube enhancement that adds time calculations (or time views) to a selected hierarchy. This enhancement supports the following categories of calculations:
List item
Period to date.
Period over period growth.
Moving averages.
Parallel period comparisons.
The wizard will let you chose the calculations and measures you want to apply.
Visit : https://learn.microsoft.com/en-us/analysis-services/multidimensional-models/define-time-intelligence-calculations-using-the-business-intelligence-wizard
Visit : http://www.ssas-info.com/analysis-services-articles/62-design/2465-ssas-time-intelligence-wizard
2- Use a dimension table to calculate your calculations, this solution is more complicated, but very powerful and one of the best practices.
The first step is to create a new physical dimension, with real
members for each of the calculations we're going to need. We don't
actually need to create a table in our data warehouse for this
purpose, we can do this with an SQL view like this
CREATE VIEW DateTool AS SELECT ID_Calc = 1, Calc = 'Real Value' UNION ALL SELECT ID_Calc = 2, Calc = 'Year To Date'
Next, we need to add this view to our DSV and create a dimension based
on it. The dimension must have one hierarchy and this hierarchy must
have its IsAggregatable property set to False. The DefaultMember
property of this hierarchy should then be set to the Real Value
member. Giving this dimension a name can be quite difficult, as it
should be something that helps the users understand what it does –
here we've called it Date Tool. It needs no relationship to any
measure group at all to work.
Our next task is to overwrite the value returned by each member so
that they return the calculations we want. We can do this using a
simple SCOPE statement in the MDX Script of the cube:
this code let you create the YEAR-TO-DATE aggregation for all your measures.
SCOPE ([Date Tool].[Calculation].[Year To Date]); THIS = AGGREGATE ( YTD ([Date Order].[Calendar].CurrentMember), [Date Tool].[Calculation].[Real Value]); END SCOPE;
Visit:https://subscription.packtpub.com/book/big_data_and_business_intelligence/9781849689908/6/ch06lvl1sec35/calculation-dimensions
I have a simple Data Cube for computing shared expenditures with one measure Amount and some dimensions and hierarchies but interesting one would be Relationship. It describes who bought something for whom. It's structure is:
Who
For whom
Relationship key
I am trying to code a calculation representing debt. For example if I bought something for sharing usage it would be half of 0.5 * Amount. On the other hand, if I bought something for myself it would be 0 * Amount.
As far I tried following calculation:
IIF(
[dimRelationship].[Relationship].currentMember = [dimRelationship].[Relationship].[RelatonshipID].&[MeShared],
[Measures].[Amount]*0.5,
[Measures].[Amount]*0)
It works good only at lowest RelationshipID level. When I roll-up browsered pivot-table it is acting according to else-expression. That is not really surprising because hierarchy's currentMember is not MeShared anymore. Another bad thing that total aggregation work neither - it would be most important as a general summary. Is there any suffix like .LeafMember or something like that which could help me perform this calculation?
Thank you in advance!
Best regards,
Max!
In you fact table add another column, in this column store the "Amount" multiplied by the relationshipID for that particular row. This will address you issue right out the box
Try SCOPE statement.
Firstly, create new calculation [Debt Calculation] as [Measures].[Amount]*0 /* as ELSE-scenario */
Then create a SCOPE:
SCOPE([Debt Calculation],[dimRelationship].[Relationship].[RelatonshipID].&[MeShared]);
THIS=[Measures].[Amount]*0.5;
END SCOPE;
If it's the lowest level of hierarchy and is a dimension key (which is used as a link to measure groups), it will re-calc higher levels of this dimension automatically. Please post a result here if not.
(This is a mock of my actual setup to help me figure out the problem.)
I have one fact table and one dimension table, linked by an id field.
My goal is to make a measure that sums up all "thing_count" (integer) values in my cube.
If the user splits by nothing, it should show the total "thing_count" for all records in the fact table. If it's split by "category_name" from the dimension, it should show the total "thing_count" for each category.
I tried to achieve this by creating a SUM measure in my cube:
It works, but not in the way I intend it to
It always shows (null) unless I drag in the "id" field from the dimension.
Measure only:
Measure and category:
Measure, category, and id:
How can I make the measure show the value without keys needing to be present?
Edit:
For GregGalloway's request (I've edited the names so the screenshots are easier to follow):
One common explanation for this behavior (no aggregation) is that you have inadvertently commented out the CALCULATE; statement in your MDX script in the cube. Please check that statement is still present.
I am new to SSAS. I have a requirement, I need to calculate no of working days between user selected date range (either in Excel or SSRS or PowerBI). I found the MDX query, I need assistance with create a named calculation with MDX expression.
Date Dimension (Filtered):
MDX:
WITH MEMBER Measures.WorkingDays AS
COUNT
(
exists( EXISTING {[Dim Date].[Date].[Date].members}
, [Dim Date].[Is Weekday].&[1] )
)
Select {Measures.WorkingDays} on 0 ,
[Dim Date].[Month].[Month] on 1
from [Project Cube]
where ([Dim Date].[Date].&[2018-01-01T00:00:00]:[Dim Date].[Date].&[2018-04-25T00:00:00])
I need to add this named column on Fact table as measurement. I am having trouble with the below items:
Creating named query with MDX expression mentioned.
Adding a [Number of Working Days] as measure in Fact table.
Please correct me, If I am doing it in wrong way. My requirement is I need a [NoOfWorkingDays] as measure in fact table, so that I can use SSAS aggregate to use it as input on other measure, such as ([utilization%] = ([ActualDaysWorked] / [NoofWorkingDays]).
Note that, I can do analysis with the given MDX, but I need to deploy it with precalculated values in cube, so that end user can directly use the cube.
Kindly let me know, if more details required, Thank you.
Welcome to SSAS and MDX. Now to the answer.
I need to add this named column on Fact table as measurement. I am
having trouble with the below items:
Creating named query with MDX expression mentioned. Adding a [Number
of Working Days] as measure in Fact table.
You dont need to add it to the Fact table at all. Open your SSAS project, in your object explorer double click your cube. Now on the top left hand you will see a CALCULATIONS tab. In the CALCULATION tab, Click new calculated member, the icon has a calculator on it.
Please correct me, If I am doing it in wrong way. My requirement is I
need a [NoOfWorkingDays] as measure in fact table, so that I can use
SSAS aggregate to use it as input on other measure, such as
([utilization%] = ([ActualDaysWorked] / [NoofWorkingDays]).
If I remember correctly, the calculated members will not be added into the Aggregations, however the underlying measures would be. Secondly if you are wondering that you can use your calculated Measure in another calculated measure. The answer is yes you can use it in another calculated measure. So this is totally possible
> ([utilization%] = ([ActualDaysWorked] / [NoofWorkingDays])
where [utilization%] and [NoofWorkingDays] are calculated measures.
I am having some issues trying to implement an average of a dimension attribute.
The basic structure is:
Booking Header Dimension
Fact Table (multiple rows per Booking Header
entry)
On the booking header dimension I have a numerical attribute called Booking Window, and I want to be able to create a calculated measure that averages this value.
We are using SQL Server 2012 standard edition.
Any help would be greatly appreciated.
The best approach would be to create a measure group from the dimension table (in BIDS, go to cube designer, tab "Cube Structure", right-click the cube object in the Measures list, and select "New Measure Group", select your dimension table). BIDS will generate some measures, and you can remove all but two: the one based on your numeric attribute (I will call it "YourSummedAttrib" to have a name to refer to below), and the count measure. The aggregate function for the measure "YourSummedAttrib" will probably be "sum", leave that as it is.
Then, create a calculated measure which divides "YourSummedAttrib" by the count measure, which gives the average. Finally, if you have tested everything, make the two measures "YourSummedAttrib" and the count measure invisible before you give the cube to the users, as they only need to see the average, which is the calculated measure.
You can try this which should give you the average of that attribute across all members.
WITH MEMBER [Measures].[Booking Window Value] AS
[Booking Header].[Booking Window].CURRENTMEMBER.MEMBER_VALUE
MEMBER [Measures].[Avg Booking Window Value] AS
AVG([Booking Header].[Booking Window].[Booking Window].MEMBERS,[Measures].[Booking Window Value])
SELECT
[Measures].[Avg Booking Window Value] ON COLUMNS
FROM
[YourCube]
Hope that helps and apologies for any confusion on my part.
Ash
I tried to use the same idea, but without success. The solution I found was create a view with the calculated average and include a new group of measures.