I'm trying to see how useful the Covariance function is in MDX. I have two measures, GP and order age. I'd like to compare the correlation between these two variables for every sales location. My current query returns the same value for all locations:
with
member [Measures].[r] as correlation( [Location].[Location Name].[Location Name].members, [Measures].[Order Age], [measures].[gp])
select [Measures].[r] on 0,
[Location].[Location Name].[Location Name].members on 1
from [DM OIP]
Where am I going wrong?
Your MDX is calculating the covariance of those two measures across all locations for every row. How about calculating the covariance of those two measures across months for the current location (the location on that row)?
with
member [Measures].[r] as correlation( [Date].[Month].[Month].members, [Measures].[Order Age], [measures].[gp])
select [Measures].[r] on 0,
[Location].[Location Name].[Location Name].members on 1
from [DM OIP]
Looks like you're missing a relationship.
So location is not related to one, or both of the measures used.
Related
I am trying to learn MDX. I am an experienced SQL Developer.
I am trying to find an example of an MDX query that has more than two dimensions. Every single webpage that talks about MDX provides simple two dimensional examples link this:
select
{[Measures].[Sales Amount]} on columns,
Customer.fullname.members on rows
from [Adventure Works DW2012]
I am looking for examples that use the following aliases: PAGES (third dimension?), section (forth dimension?) and Chapter (fifth dimension?). I have tried this but I do not think it is correct:
select
{[Measures].[Sales Amount]} on columns,
Customer.fullname.members on rows,
customer.Location.[Customer Geography] as pages
from [Adventure Works DW2012]
I am trying to get this output using an MDX query (this is from AdventureWorks DW2012):
That's not a 3-dimensional resultset in your screenshot, unless there's something cropped from it.
Something like
SELECT [Geography].[Country].Members ON 0,
[Customer].[CustomerName].Members ON 1
FROM [whatever the cube is called]
WHERE [Measures].[Sales Amount]
(dimension/hierarchy/level names may not be exactly right)
would give a resultset like the one in your message.
The beyond 2nd-dimension dimensions and dimension names are not used in any client tool that I know. (Others may know different). They seem to be there in MDX so that MDX can hand >2-dimensional resultsets to clients that can handle them (e.g. an MDX subquery handing its results to the main query).
An often-used trick in MDX is to get the members of two dimensions onto one axis by cross-joining:
SELECT
{[Date].[Calendar Date].[Calendar Year].Members * [Geography].[Country].Members} ON 0,
[something else] ON 1
FROM [Cube]
How about the following - it does not send more than two dimensions back to a flat screen but it uses quite a few dimensions explicitly:
SELECT
[Measures].[Sales Amount] ON O,
[Customer].[fullname].MEMBERS ON 1
FROM
(
SELECT
[Date].[Calendar Month].[Calendar Month].&[February-2012] ON 0,
[Geography].[Country].[Country].&[Canada] ON 1,
[Product].[Product].&[Red Bike] ON 2,
[Customer].[Customer].&[foo bar] ON 3
FROM [Adventure Works DW2012]
)
I've made up the dimension | hierarchy | member combinations as I do not have access to the cube.
Also if we consider implicit dimensions then take the following:
SELECT
[Customer].[Location].[Customer Geography] ON 0,
[Customer].[fullname].[fullname].&[Aaron Flores] ON 1
FROM [Adventure Works DW2012]
WHERE
(
[Measures].[Sales Amount]
);
On the slicer I've used braces (..) which indicate a tuple, but this is actually shorthand for the following:
SELECT
[Customer].[Location].[Customer Geography] ON 0,
[Customer].[fullname].[fullname].&[Aaron Flores] ON 1
FROM [Adventure Works DW2012]
WHERE
(
[Measures].[Sales Amount]
,[Date].[Calendar Month].[Calendar Month].[All],
,[Geography].[Country].[Country].[All],
,[Product].[Product].[All]
,...
,...
....
);
The All member from every dimension in the cube could be included in this slicer without affecting the result.
So the whole nature of mdx is multi-dimensional - yes you do not get more than a 2 dimensional table returned to your screen but the way you get to that cellset could well involve many dimensions.
The following script gives exactly the result I want.
It feels like a hack as I've added the custom members VALUE and VALUE_MTD onto the hierarchy [Customer].[Country]. I've chosen this hierarchy arbitrarily - just not used [Measures] or [Date].[Calendar] as they are already in use.
Is there a more standard approach to returning exactly the same set of cells?
WITH
MEMBER [Customer].[Country].[VALUE] AS
Aggregate([Customer].[Country].[(All)].MEMBERS)
MEMBER [Customer].[Country].[VALUE_MTD] AS
Aggregate
(
PeriodsToDate
(
[Date].[Calendar].[Month]
,[Date].[Calendar].CurrentMember
)
,[Customer].[Country].[VALUE]
)
SELECT
{
[Customer].[Country].[VALUE]
,[Customer].[Country].[VALUE_MTD]
} ON 0
,NON EMPTY
{
[Measures].[Internet Sales Amount]
,[Measures].[Internet Order Quantity]
}
*
Descendants
(
{
[Date].[Calendar].[Month].&[2007]&[12]
:
[Date].[Calendar].[Month].&[2008]&[01]
}
,[Date].[Calendar].[Date]
) ON 1
FROM [Adventure Works];
The standard approach is called utility dimension. If you Google this term, you will find several descriptions of this approach. A "utility dimension" is one which does not reference any data, but is just added to the cube for the purpose of being able to cross join them with all other dimensions for calculations. You can have one or more of them.
Thus, in most cases, physically there is nothing in the dimension. It is just used for calculated members. (Depending on the implementation, you may have the attribute members defined physically, if you want to have some properties for them. But then, only the default member is referenced in the star schema from the fact tables. The attribute member values are then overwritten in the calculation script.)
Typical applications for this are time calculations like YTD, MTD, MAT (Moving Annual Total, i. e. a full year of data ending in the selected date), or comparisons like growth vs. a previous period.
I am using Performance Point Dashboard Designer 2013 and SharePoint Server 2013 for building dashboards. I am using SSAS2012 for Cube.
I have a scenario similar to the one illustrated by figure below. I am required to find Previous Non-Empty value for purpose of finding Trends.
Measure: [Quota]
Dimension: [Date].[Calendar Date].[Date]
The script ([Measures].[Quota], [Date].[Calendar Date].PrevMember) gives you a previous date. Lets say for date 27-Jan-13 whose Quota value is 87, it returns 26-Jan-13 which has null value. I want it to return 21-Jan-13 that has some Quota value. And for date 21-Jan-13, I want to return 15-Jan-13.
I wonder if this is possible.
Thanks,
Merin
After long searches and hits & trials and so on, I think I invented a solution of my own for myself.
Following is the script for my Calculated Member.
(
[Quota],
Tail
(
Nonempty
( LastPeriods(15, [Date].[Calendar Date].PrevMember)
,[Quota]
)
).Item(0)
)
Explanation
The number 15 means it will look for non-empty measures up to 15 siblings.
Now we know up to how many siblings to traverse back, in this case 15.
Lets find 15 previous siblings (both empty and non-empty) excluding current member.
(LastPeriods(15, [Date].[Calendar Date].PrevMember)
Since it will yield both empty and non-empty members, lets filter out empty members in terms of measure [Quota]. If we don't specify measure here, it will use default measure whatever it is and we may not get desired result.
Nonempty(LastPeriods(15, [Date].[Calendar Date].PrevMember),[Quota])
We may have several members in the output. And we will choose the last one.
Tail
(
Nonempty
( LastPeriods(15, [Date].[Calendar Date].PrevMember)
,[Quota]
)
)
So far, the script above gives previous non-empty member. Now we want to implement this member for our measure [Quota].
Hence we get the script below ready to create a Calculated Member.
(
[Quota],
Tail
(
Nonempty
( LastPeriods(15, [Date].[Calendar Date].PrevMember)
,[Quota]
)
).Item(0)
)
You can use recursion to define this.
The following query delivers something similar for the Adventure Works cube:
WITH member [Measures].[Prev non empty] AS
IIf(IsEmpty(([Date].[Calendar].CurrentMember.PrevMember, [Measures].[Internet Sales Amount])),
([Date].[Calendar].CurrentMember.PrevMember, [Measures].[Prev non empty]),
([Date].[Calendar].CurrentMember.PrevMember, [Measures].[Internet Sales Amount])
), format_String = '$#,##0.00'
SELECT {[Measures].[Internet Sales Amount], [Measures].[Prev non empty]}
ON COLUMNS,
non empty
Descendants([Date].[Calendar].[Month].&[2007]&[12], [Date].[Calendar].[Date])
ON ROWS
FROM [Adventure Works]
WHERE [Customer].[Customer].&[12650]
You would have to replace the name of the date hierarchy, as well as the measure name from Internet Sales Amount to Quota in the recursive definition of the measure Prev non empty.
I am trying to calculate percentile (for example 90th percentile point of my measure) in a cube and I think I am almost there. The problem I am facing is, I am able to return the row number of the 90th percentile, but do not know how to get my measure.
With
Member [Measures].[cnt] as
Count(NonEmpty(
-- dimensions to find percentile on (the same should be repeated again
[Calendar].[Hierarchy].members *
[Region Dim].[Region].members *
[Product Dim].[Product].members
,
-- add the measure to group
[Measures].[Profit]))
-- define percentile
Member [Measures].[Percentile] as 90
Member [Measures].[PercentileInt] as Int((([Measures].[cnt]) * [Measures].[Percentile]) / 100)
**-- this part finds the tuple from the set based on the index of the percentile point and I am using the item(index) to get the necessary info from tuple and I am unable to get the measure part
Member [Measures].[PercentileLo] as
(
Order(
NonEmpty(
[Calendar].[Hierarchy].members *
[Region Dim].[Region].members *
[Product Dim].[Product].members,
[Measures].[Profit]),
[Measures].[Profit].Value, BDESC)).Item([Measures].[PercentileInt]).Item(3)**
select
{
[Measures].[cnt],
[Measures].[Percentile],[Measures].[PercentileInt],
[Measures].[PercentileLo],
[Measures].[Profit]
}
on 0
from
[TestData]
I think there must a way to get measure of a tuple found through index of a set. Please help, let me know if you need any more information. Thanks!
You should extract the tuple at position [Measures].[PercentileInt] from your set and add the measure to it to build a tuple of four elements. Then you want to return its value as the measure PercentileLo, i. e. define
Member [Measures].[PercentileLo] as
(
[Measures].[Profit],
Order(
NonEmpty(
[Calendar].[Hierarchy].members *
[Region Dim].[Region].members *
[Product Dim].[Product].members,
[Measures].[Profit]),
[Measures].[Profit], BDESC)).Item([Measures].[PercentileInt])
)
The way you implemented it, you tried to extract the fourth (as Item() starts counting from zero) item from a tuple containing only three elements. Your ordered set only has three hierarchies.
Just another unrelated remark: I think you should avoid using complete hierarchies for [Calendar].[Hierarchy].members, [Region Dim].[Region].members, and [Product Dim].[Product].members. Your code looks like you are including all levels (including the all member) in the calculation. But I do not know the structure and names of your cube, hence I may be wrong with this.
An alternate method could be to find the median of the last 20% of the records in the table. I've used this combination of functions to find the 75th percentile. By dividing the record count by 5, you can use the TopCount function to return a set of tuples that make up 20% of the whole table sorted in descending order by your target measure. The median function should then land you at the correct 90th percentile value without having to find the record's coordinates. In my own use, I use the same measure for the last parameter in both the Median and TopCount functions.
Here's my code:
WITH MEMBER Measures.[90th Percentile] AS MEDIAN(
TOPCOUNT(
[set definition]
,Measures.[Fact Table Record Count] / 5
,Measures.[Value by which to sort the set so the first 20% of records are chosen]
)
,Measures.[Value from which the median should be determined]
)
Based on what you've supplied in your problem definition, I would expect your code to look something like this:
WITH MEMBER Measures.[90th Percentile] AS MEDIAN(
TOPCOUNT(
{
[Calendar].[Hierarchy].members *
[Region Dim].[Region].members *
[Product Dim].[Product].members
}
,Measures.[Fact Table Record Count] / 5
,[Measures].[Profit]
)
,[Measures].[Profit]
)
I'm trying to write a query to give me the total number of users for each customer per day.
Here is what I have so far, which for each customer/day combination is giving the total number of user dimension entries without splitting them up by customer/day.
WITH MEMBER [Measures].[MyUserCount]
AS COUNT(Descendants([User].CurrentMember, [User].[User Name]), INCLUDEEMPTY)
SELECT
NON EMPTY CrossJoin([Date].[Date].Members, [Customer].[Customer Name].Members) ON ROWS,
{[Measures].[MyUserCount]} on COLUMNS
FROM
[Users]
The problem with your calculated member is that [User].CurrentMember is set to the All member for every row tuple, and thus the count is the total. What you need is a way for the [Customer].CurrentMember and [Date].CurrentMember to effectively filter the [User] dimension.
You need to use a measure that makes sense, i.e. that will have a non-empty value for meaningful joins of the dimension members that you're interested in.
To find this out, you could start by running a query like this:
SELECT
NON EMPTY CrossJoin(
[User].[User Name].Members,
[Measures].[Some measuse]
) ON COLUMNS,
NON EMPTY CrossJoin(
[Date].[Date].Members,
[Customer].[Customer Name].Members
) ON ROWS
FROM [Project]
You would have selected Some measure adequately. The results of that query will be a lot of empty cells, but in a given row, the columns that do have a value correspond to the Users that are related to a given Customer x Date tuple (on the row). You want to count those columns for every row. COUNT and FILTER are what you need, then the query with the calculated member will be
WITH MEMBER [Measures].[User count] AS
COUNT(
FILTER(
[User].[User Name].Members,
NOT ISEMPTY([Measures].[Some measure])
)
)
SELECT
NON EMPTY {[Measures].[User count]} ON COLUMNS,
NON EMPTY CrossJoin(
[Date].[Date].Members,
[Customer].[Customer Name].Members
) ON ROWS
FROM [Users]
I am assuming a fair bit here, but with some experimentation you should be able to work it out.