I'm having a bit of trouble accomplishing something that I think should be relatively straightforward in MDX. I would like to create a calculated member that provides a sum of one of my measures over the previous two weeks at a given point in time. My time dimension looks like:
TimeId TradingDate Day of Week
-----------------------------------
1000 11/1/2012 Thursday
1001 11/2/2012 Friday
1002 11/5/2012 Monday
1003 11/6/2012 Tuesday
... ...
What makes this particularly difficult is that my Time dimension is not quite complete. The members of my Time dimension only correspond to trading days in the stock market, and not all time. This means that weekends, holidays, or any other day in which the stock market is closed are excluded. This also means the normal methods of traversing time such as LAG or PARALLELPERIOD will not work quite right here. LAG(14), for example, means "14 trading days", which at any given point could represent a variable length of actual time.
Inside my calculated member, I'm attempting to use FILTER in order to get only time members that are within the previous two weeks of the CurrentMember. However, I can't seem to figure out the proper syntax (if there is one) to accomplish this. I imagine it would be something like:
WITH MEMBER [Sum of Price Previous 2 Weeks] AS
SUM(
FILTER(
[Time].[TimeId].Children
, [Time].[TradingDate].MemberValue
>= VBA!DATEADD("ww", -2, [Time].[TradingDate].CurrentMember.MemberValue)
)
, [Price]
)
However, this doesn't quite work. I can't seem to separate the context of the calculated members current iteration from what would be a separate context inside of the FILTER function. In other words, I'm not sure how to say:
"When iterating over the set inside of FILTER, compare the current
member of each iteration against the value of the CurrentMember in
the scope of the calculated member"
Is what I'm trying to accomplish even possible? Is there a different approach I could be taking to accomplish what I'm after?
The result you'll get from a calculated member will depend on the axis of your query. So first, make sure you have [Time].[TradingDate] in your axis.
Second, your [Time].[TradingDate] hierarchy should be ordered by Key (I assume TradingDate is the key).
Now you can use this member definition:
WITH MEMBER [Sum of Price Previous 2 Weeks] AS
SUM(
[Time].[TradingDate].CurrentMember.Lag(14):[Time].[TradingDate].CurrentMember, [Price]
)
You can use set aliases to refer to the outer CurrentMember in the Filter context:
WITH MEMBER [Sum of Price Previous 2 Weeks] AS
SUM(
GENERATE([Time].[TradingDate].CurrentMember AS CurrentDateAlias,
FILTER(
[Time].[TimeId].Children
, [Time].[TradingDate].MemberValue
>= VBA!DATEADD("ww", -2, CurrentDateAlias.Item(0).MemberValue)
)
)
, [Price]
)
GENERATE is used just to define the alias somewhere.
Related
Would much appreciate any help on this.
I have a measure called "Sales" populated with values, however i am trying to turn the "Sales" value to 0, whenever the "Sales Flag" is set to 0.
Important Note: The Sales Flag is based on Date (lowest level of detail).
The difficulty that i am really experiencing and cant get a grip on, is how i am trying the display the MDX outcome.
As explained above, i would want to make the "Sales" value 0 whenever we have a 0 in the "Sales Flag" (which is based on the Date), but when I run the MDX Script I would wan't on the ROWS to NOT display the Date, but instead just the Week (higher Level to Date), as below shows:
I really have spent hours on this and can't seem to understand how we can create this needed custom Sales measure based on the Sales Flag on the date level, but having the MDX outcome display ROWS on Week level.
Laz
You need to define the member in the MDX before the select. Something like that:
WITH MEMBER [Measures].[Fixed Sales] as IIF([Sales Flag].currentMember=1,[Sales], 0)
SELECT [Measures].[Fixed Sales] on 0, [Sales Flag] on 1 from [Cube]
I am writing the code without SSAS here so it might not be the 100% correct syntax but you can get the general idea ;)
You can add the iif in the SELECT part but I find creating member to be the cleaner solution.
SELECT IIF([Sales Flag].currentMember=1,[Sales], 0) on 0, [Sales Flag] on 1 from [Cube]
If you have a control over the cube in SSAS you can create a calculated member there and you can access it easier.
Glad to hear if Veselin's answer works for you, but if not...
Several approaches are also possible.
Use Measure expression for Sales measure:
Use SCOPE command for Day level (if it's Key level of Date dimension). If it's not a key level you have to aggregate on EVERY level (week, year etc) to emulate AggregateFunction of Sales measure but with updated behavior for one flag:
SCOPE([Date].[Your Date Hierarchy].[Day].members,[Measures].[Sales]);
THIS=IIF([Sales Flag].CurrentMember = 1,[Measures].[Sales],0);
END SCOPE;
Update logic in DSV to multiply Sales column by SalesFlag. This is the easiest way from T-SQL perspective.
I'm trying to add a calculated member to my cube, which will return the first fiscal year where there is any data at all in a particular measure.
The purpose is to suppress (i.e. NULLify) various year-on-year calculated measures when the year is this first year: in that year, comparison with the previous year is meaningless.
I've got this so far:
WITH MEMBER Measures.DataStartYear_Sales
AS
HEAD(
NONEMPTY([Calendar].[Fiscal Periods].[Fiscal Year].Members,[Measures].[QuantityOrdered])
,1).Item(0).Properties("NAME")
At the moment:
a. It's a query-scoped measure, as that's easier to experiment with.
b. It returns the first year's Name, as that's easier to see. Eventually I'll just return the member itself, and do an IS comparison against the year hierarchy .CurrentMember in the other calculated member calculations.
The problem I expected, which has happened, is that I only want this measure to be calculated once, over the whole cube. But when I used it in a query, it obviously reacts to the context. For example, if I stick the Products dimension on ROWS, the value of this measure will be different for each row, because each product's earliest order date is different.
That is of course useful, but it's not what I want. Is there some way to force this measure to ignore the query context, and always return the same value?
I looked into SCOPE_ISOLATION and SOLVE_ORDER, but they don't do what I'm trying to do here.
I suppose I could specify a tuple of Dimension1.All, Dimension2.All.... DimensionN.All, covering all dimensions in the cube, but that seems messy and fragile.
I think you might be able to accomplish this with static sets. Here is an example using Adventure Works that produces the same first year regardless of context:
WITH STATIC SET FirstYear AS
HEAD
(
NONEMPTY([Date].[Calendar Year].[Calendar Year].MEMBERS, [Measures].[Internet Sales Amount])
, 1
)
MEMBER FirstYearName AS
FirstYear.ITEM(0).NAME
SELECT
[Measures].[FirstYearName] ON COLUMNS
, [Date].[Calendar Year].[Calendar Year].MEMBERS
//Add as many dimensions as you like here...for example
* [Product].[Product].[Product].MEMBERS
ON ROWS
FROM
[Adventure Works]
;
Example output:
That should hopefully put you on the right track.
I have a multidimensional cube that needs a custom measure that I'm not sure how to build.
That data looks like this:
MemberID-----Date-------EventType
1--------------1/1/2016-------1
2--------------1/1/2016-------2
3--------------2/1/2016-------1
2--------------2/1/2016-------2
4--------------2/1/2016-------2
There is a count measure in the cube, but others can be added if needed. I need to create a measure that will use whatever filters the user applies and then count the EventType (1 and 2 only) by month, divide the resulting counts for EventType 1 into the count for EventType 2 (for each month individually), and finally sum the monthly results. For example 1/1/2016 would be 1/1=1 (count of EventType 1 and count of EventType 2) and 2/1/2016 would be 1/2=0.5 so the resulting measure value for the two months would be 1+0.5=1.5. Any help is greatly appreciated.
Let's assume you have a Date dimension with an attribute called Month. And let's assume you have an EventType dimension. And let's assume you have a count measure in your measure group called Cnt. Here's what else you need to do.
First, go to the DSV and add a new calculated column to the fact table which is called NullInt and is the following expression:
cast(null as int)
Then create a new Sum measure in your measure group off that column and call the measure My Rollup. Under the Source property, change NullHandling to Preserve so that it will start off null.
To explain why we're doing this, a scoped assignment to a physical measure will aggregate up. (If you assign a value to a physical measure at the grain of each month, then it will rollup to the grand total.) But a scoped assignment to a calculated measure doesn't roll up.
Then in your MDX script add the following calculations:
scope([Date].[Month].[Month].Members); //calculate at the month level then rollup
[Measures].[My Rollup] = DIVIDE(
([Event Type].[Event Type].&[1],[Measures].[Cnt]),
([Event Type].[Event Type].&[2],[Measures].[Cnt])
);
end scope;
Note that your version of SSAS probably has the DIVIDE function if it's AS2012 with the latest service pack or newer. But if it doesn't, you can always do division the old fashioned way as IIF(denom=0,null,num/denom).
I would like to calculate Sum(QTY) until the start date of the month for a given date.
Basically I can calculate Sum(QTY) until given date in my measure like:
SumQTYTillDate:=CALCULATE(SUM([QTY]);FILTER(ALL(DimDateView[Date]);DimDateView[Date]<=MIN(DimDateView[Date])))
But I also would like to calculate Sum(QTY) for dates before 10/1/2015 - which is the first date of selected Date's month. I have changed above measure and used STARTOFMONTH function to find first day of the month for a given date like;
.......DimDateView[Date]<=STARTOFMONTH(MIN(DimDateView[Date]))))
but not avail, it gives me
"A function ‘MIN’ has been used in a True/False expression that is
used as a table filter expression. This is not allowed."
What am I missing? How can I use STARTOFMONTH function in my measure?
Thanks.
STARTOFMONTH() must take a reference to a column of type Date/Time. MIN() is a scalar value, not a column reference. Additionally, your measure wouldn't work, because STARTOFMONTH() is evaluated in the row context of your FILTER(). The upshot of all this is that you would get a measure which just sums [QTY] across the first of every month in your data.
The built in time intelligence functions tend to be unintuitive at best. I always suggest using your model and an appropriate FILTER() to get to what you want.
In your case, I'm not entirely sure what you're looking for, but I think you want the sum of [QTY] for all time before the start of the month that the date you've selected falls in. In this case it's really easy to do. Add a field to your date dimension, [MonthStartDate], which holds, for every date in the table, the date of the start of the month it falls in. Now you can write a measure as follows:
SumQTY=SUM(FactQTY[QTY])
SumQTYTilStartOfMonth:=
CALCULATE(
[SumQTY]
;FILTER(
ALL(DimDateView)
;DimDateView[Date] < MIN(DimDateView[MonthStartDate])
)
)
I'd like an easy way of getting the total number of days for each month in the date dimension.
Currently this information is not exposed in our cubes. Therefore I need to write custom mdx such as the following:
WITH
SET [13Mth] AS Tail([Date].[Date - Calendar Month].[Calendar Month].MEMBERS ,13)
SET [m] AS Tail([13Mth])
MEMBER [Measures].[TotalNumDaysMth] AS
Datepart
("D",
Dateadd
("M",1,
Cdate(Cstr(VBA!Month([m].Item(0).Item(0).Lag(1).Name)) + "-01-" + Cstr(VBA!Year([m].Item(0).Item(0).Lag(1).Name)))
)
- 1
)
MEMBER [Measures].[TotalNumDaysMth-1] AS
Datepart
("D",
Dateadd
("M",1,
Cdate(Cstr(VBA!Month([m].Item(0).Item(0).Lag(1).Name)) + "-01-" + Cstr(VBA!Year([m].Item(0).Item(0).Lag(1).Name)))
)
- 1
)
I don't believe our users will need this information within our cube browsing client but from a developer point of view I could do without having to always implement the above.
What approach should we use to make the above data more easily available?
I have always added an attribute to the date dimension "days in month", of type integer. You can hide this attribute, if you do not want to expose the attribute hierarchy to your users.
But you can still use in calculations.
So my advise would be add a proper attribute to your time dimension and base your calculations of on the attributes.
Hope this helps somehow.
I am assuming that in your development scenario, the month will come from Web Service/Front End/SSRS report etc. In any such cases, you can create a measure that will return the count.
Approach 1: Year-Quarter-Month-Date exists
It is a good practice to have a similar hierarchy in place. The following MDX will work:
with set CurrentMonth as
[Date].[Year-Quarter-Month-Date].currentmember
set DatesInCurrentMonth as
descendants(abc, [Date].[Year-Quarter-Month-Date].[Date])
member [Measures].countofdaysinmonth as
count(DatesInCurrentMonth)
select measures.CountOfDaysInMonth on 0
from [MyCube]
where [Date].[Year-Quarter-Month-Date].[Month].&[Feb-2004]
Approach 2: Such hierarchy doesn't exist
with set CurrentMonth as
[Date].[Month].currentmember
set DatesInCurrentMonth as
exists([Date].[Date].children, abc, "<<Any Measure group>>")
member measures.countofdaysinmonth as
count(DatesInCurrentMonth)
select measures.countofdaysinmonth on 0
from [MyCube]
where [Accident Date].[Month].&[Feb-2004]
Let me know if they work/have issues.