Cascade Filter for Hierarchy dimension

Hi experts, I try to develop a design Studio Dashboard with BW DS. I would like to create a cascade filter for the same dimension but in different levels from hierarchy. First filter would have upper level node of the dimension that will filter same dimension but in lower level values. First selector has manually added Items corresponding the Hierarchy nodes from Level 2. I didn't find how to get specific level of dimension node with function getMemberList(). I created a DS with Level 2 of hierarchy but when I get the values with DS.getMemberList("ZREP", MemberPresentation.INTERNAL_KEY, MemberDisplay.TEXT, "ALL")) I get all final values from last level node. There is a way to get specific values from hierarchy level with getMemberList() ? Once I get all list values from a level on REP_SELECTOR on click on the selector has the following script: DS_1.setFilter("ZREP", REP_SELECTOR.getSelectedValue()); SHOP_SELECTOR.setItems(DS_1.getMemberList("ZREP", MemberPresentation.INTERNAL_KEY, MemberDisplay.TEXT, "ALL")); Unfortunately shop_selector is showing all values from ZREP without any previous filter. Do you know what is wrong on it? Thank you very much

Hi Oscar,
So let me explain a solution that will hopefully meet your needs. A couple of comments before we begin. Since you need to cascade through a hierarchy, for your first list selector it is not possible to use the standard Dropdown Box or List Box components because as you know, these must be populated with the getMemberList() method and this method does not return the hierarchy structure but only the member values (excluding hierarchy nodes). So my proposed solution relies on using a CrossTab component as a selection list. The benefit of this approach is that the hierarchy structure is preserved.
The application layout is as follows:
This example assumes that the selection in the first filter list populates (cascades) the second filter list and the second filter list selection filters the chart. The first filter list selection does not also filter the chart but this functionality can easily be added if necessary.
The steps are as follows:
1. The hierarchy has been set up as per your example, with a topmost root node and one level of text nodes below that. The final level is represented by characteristics. A screenshot is shown below:
2. Define two global script variables to represent the key value of the root node of the hierarchy as well as the standard member selection value for all members, as shown below:
3. Add 3 data sources to the application, all based on the same BEx Query, which includes the dimension (characteristic) that has the hierarchy, as shown below:
4. Add two CrossTab components and one Chart component as shown above.
5. Activate the hierarchy for the first data source DS_1 and set the default level to 2 as shown below:
6. Activate the hierarchy for the second data source DS_2 and set the default level to 2 as shown below:
7. Set the default members of the second data source DS_2 to represent the first level 1 group in the hierarchy as shown below. This is required to ensure that the second list is populated with a default selection on startup.
7. Define the initial view of the third data source DS_3 in accordance with the requirements of the chart, as follows:
8. Assign data source DS_1 to the first CrossTab, assign data source DS_2 to the second CrossTab and assign the third data source DS_3 to the Chart.
9. Define the script for the On Select event of the first CrossTab as shown below. The purpose of this script is to populate the second CrossTab based on the selection in the first one to cascade the filter. Note that a selection of the root node is excluded to prevent duplication of the first level of the hierarchy in the second CrossTab list.
10. Define the script for the On Select event of the second CrossTab as shown below to filter the chart based on the selection:
And that's it! Let me know if this gets us closer to what you're looking for. We can then refine the approach as needed.
Regards,
Mustafa.

Similar Messages

Filter for one hierarchy and expanding for another one?

Hi to all,
We've a dimension A with 2 different hierarchies (call them as usual PARENTH1 and PARENTH2).
What I've been asked by the users is a little interface, where they can:
- filter for one parent of the first hierarchy
- expand for the parents of the second hierarchy, who share the same base members.
Initially I had hoped that imposing the value desired of the parenth1 in the PageKeyRange, and expanding with a filter like HIR="H2" and HLEVEL="2" would work, but it doesn't work (I get ALL the parents at level 2 of the second hierarchy, not the parents who shares the same basemembers).
Is it possible to model this with an evDRE? If so how? Unfortunately, I cannot make a single hierarchy (this parents aren't in a 1-N relationship between them, unfortunately).
Thank you
Daniele

Hi James,
could you explain more in detail your solution? Because I tried it, and it didn't work as I expected...
Let me explain with an example
Hierarchy1:
Dir_01 has got 4 basemembers (test1, test2, test3, test4), while Dir_02 has got other 2 basemembers (test5, test6)
Hierarchy 2
Es1 has got 3 basemembers (test1, test2, test3), Es2 has got 1 basemember(test4), Es3 has got 2 basemembers (test5, test6)
What I would like to do is to
- filter an evDRE with Dir_01
- "expand" (I believe i've wrongfully explained myself in this point) with the other Hierarchy, which means: I want in the report only the nodes in the hierarchy 2 that share the same basemembers
In this way, I would have as available keys in my evDRE report:
Es1
Es2
With your solution, I even get Es3, which doesn't have any common basemember...
Kind regards
Daniele

Not possible to filter for more than one dimension in Datasource script filter method?

Hi experts.
We have a scenario where we need to apply filter on multiple dimensions .Eg.
We have one data source DS1 with Dimension HUb and material_grp .
Now we want to filter the DS1 with HUB and Material_grp at the same time.
I am using below statement to apply filter.
DS_1.setFilterExt("ZHUBCODE", "HUB1" );
DS_1.setFilterExt("ZMAT_GRP", "MATG1" ); //Clears HUB filter and shows data for all HUBS , filter with material group only
But Setfilter method first delete the last filter and apply new one.And set filter method takes only one dimension at a time.
Please help me in this

Hi Rajender,
Just as a follow-up, if you are using BW as a data source, make sure for each of the dimensions, ZHUBCODE and ZMAT_GRP, that in the Initial View Editor of the data source, the Members for Filtering > Only Values with Posted Data option is selected, which should be the default anyway.
Again if you are using BW, the other option you could try is to use the internal key for filtering with setFilter instead of setFilterExt. Is there a particular reason why you are using the external key?
Regards,
Mustafa.

Create hierarchy for a dimension in Oracle Express Administrator

I have a lookup table called 'Location'. This lookup table has a primary key PKLOC which is made up of 2 columns, State Varchar2(20) and City Varchar2(20). This primary key obviously is a lookup for a dimension of one of my cube and the data in the primary key is hierarchical with 2 levels, namely state and city. I need to import this data into my dimension, Location, using Relational import, in the following form.
eg. Level 1 Level 2
MD ---- Odenton, Baltimore, Fort Meade
VA ---- Dulles, Fair Fax
where MD and VA will come from the column State Varchar2(20) and Odenton, Baltimore, Fort Meade
Dulles and Fair Fax will come from the column City Varchar2(20).
The values in my lookup table will look like this
State City
MD Odenton
MD Baltimore
MD Fort Meade
VA Dulles
VA Fair Fax
how do I import it.

You need to do this in your source table.
The source table can look something like this
qtr_id, qtr_name, month_id, month_name
1 , Qtr1 , 101 , january
1 , Qtr1 , 102 , febuary
1 , Qtr1 , 103 , march
2 , Qtr2 , 104 , april
2 , Qtr2 , 105 , may
2 , Qtr2 , 106 , june
Then map it to the dimension, and maintain it.
Keep in mind that the member(id) fields need to be unique. I.E january 2007 and january 2006 cannot have the same id.
Ragnar

Performance Point Cascading filter challange - Can it be done? If not, we need to change our BI front end!!

My challenge at the moment is that I need to create a dynamic filter. To explain this is a cascading filter but not your normal cascading filter,
ie, Year/Month/day etc.
Here we have a single Dimension, Organisation. It contains every level of a business,
Site,
Dept,
Company name,
Division
Holding
Strategic Area
From these as a default we have 4 built hierarchies with a different combination of the above.
Activity (Strategic Area, Dept)
Organisation (company Name, site, Dept)
Sites (Division, Site)
Holdings(Holding, Strategic Area, site, Dept)
At any given time we have have additional Hierarchies created within the Dimension, further adding to the hierarchy list.
We now have a challenge where we require our users to be able to in the first filter select the Hierarchy they want to use, ie, Holdings, Site, Organisation
or Activity. This should then cascade to the second filter and contain the relevant hierarchy and levels as a multi select filter. Of course this also needs to be connected to number of graphs, charts and scorecards on the page, dynamically.
I have gone around the houses to get this to work but not having any luck. Tried the variety of filters available, MDX, member selection, tabular,
SharePoint, and attempted to mix and match to get this to work. I cannot find a way but I don't believe this is not possible.
If it is not possible we must look around for another front end solution for our BI Dashboards which I would prefer to avoid. Any response or suggestion
would be much appreciated.
Regards,
Sheb

This is possible within PerformancePoint 2010. Filters can be cascading. There is one restriction that might affect you, and that is the cascade only works on the preceding level. For example, let's say you have 4 filters, Country, County, City, postcode,
that have been set up in PerformancePoint as cascading filters. Now when you choose a country, say United Kingdom, only Counties within the UK will be displayed. That's exactly as you could expect. However, now if you select Norfolk county, all the cities
with "Norfolk" county will be available. That's including the cities within "Norfolk" Massachusetts. This is because PerformancePoint cascading filters only work between two levels.
Here's a link to setting up cascading filters http://technet.microsoft.com/en-us/library/hh272541(v=office.15).aspx
Hope that information is useful to you and the cascade limitation doesn't limit you too much.
Richard

Drill through for Attribute dimensions

Hi,
How to enable drill thrus for Attribute dimension members. I have included the Attribute dimension in Drill thru intersections, but still I am not able to do drill thrus on Attribute dimension members. Is there any other settings I need to do in Metaoutline?
Thanks
Kannan.

Yep, this is a limitation. I suggest you create two reports one that takes your "optional" dimension out and a second one that has it in. if they don't need the dimension, select the first report and if they need the filter then the second. I m always a little leery with recursive hierarchies and Gen 2 or 3 sine they create a list of level 0 members under the selection in the query and there are limits to the number of members in an in clause and at high levels in the hierarchy it is easy to surpass it (In Oracle it is 1000 members)

How to create Hierarchies for a Dimension?

Hi,
How to create Hierarchies for a Dimension in Multidimensional Models? Could you please suggest some document provides step by step instructions to create Multidimensional Model?
Thanks
M Thiyagarajan

Hi M Thiyagarajan,
connect dimension to first level and after that connect that level to another one and so on. For level - each additional outgoing connection means additional hierarchy.
Philip

Preaggregation across value based hierarchy dimension in 11g

Hi All,
I have created a cube with 6 dimensions in olap 11g. One of those six dimensions has only one hierarchy which is value based hierarchy. I have chosen level based aggregation as I know at what levels exactly users are going to query. When I was going through the dimensions to choose levels to preaggregate I noticed there were no options available for my value based hierarchy dimension ( I could see "all" option for the same case in 10g) then I tried to look for definitions of underlying objects just to make sure it will preaggregate data across my value based hierarchy dimension.
I found value set corresponding to my value based hierarchy dimension in <CUBE NAME>SOLVEAGGMAP object , which AWM uses to decide which dimension values to preaggregate but if I do rpr on that value set (rp r<CUBE NAME>SOLVE<DIMENSION NAME>_PVSET) it shows NA so my question is can I pre aggregate across value based hierarchy dimension in olap 11g?
Olap Version: 11.2.0.1
AWM version: 11.2.0.1
Thanks

Even if you know exactly which levels your users will query, percent based precompute (e.g. 30%) may still be faster in practice because queries are returned using 'sparse looping' instead of 'dense looping'. This was the single biggest performance advantage of 11g over 10g.
But if you still want to use level based precompute, then you should look at the XML template for the cube (as saved by AWM, for example). In it you should find something called PrecomputeCondition. This defines the set of members that are precomputed. Here is an example I just created using the GLOBAL schema
<PrecomputeCondition>
<![CDATA[
"TIME" LEVELS ("TIME"."MONTH", "TIME".CALENDAR_QUARTER, "TIME".CALENDAR_YEAR),
CHANNEL LEVELS (CHANNEL.TOTAL_CHANNEL, CHANNEL.CHANNEL),
CUSTOMER LEVELS (CUSTOMER.MARKET_SEGMENT, CUSTOMER.REGION, CUSTOMER.SHIP_TO),
PRODUCT LEVELS (PRODUCT.CLASS, PRODUCT.FAMILY, PRODUCT.ITEM)]]>
</PrecomputeCondition>The PrecomputeCondition is also visible through the USER_CUBES view.
SELECT PRECOMPUTE_CONDITION
FROM USER_CUBES
WHERE CUBE_NAME = 'MY_CUBE';You can hand modify this condition in the XML to specify an alternative 'non level based' precompute condition for any dimension. For example, if you define an attribute named 'SHOULD_PRECOMPUTE' on your PRODUCT dimension that is 1 for members to be precomputed and 0 for all others, then you can change the condition as follows.
<PrecomputeCondition>
<![CDATA[
"TIME" LEVELS ("TIME"."MONTH", "TIME".CALENDAR_QUARTER, "TIME".CALENDAR_YEAR),
CHANNEL LEVELS (CHANNEL.TOTAL_CHANNEL, CHANNEL.CHANNEL),
CUSTOMER LEVELS (CUSTOMER.MARKET_SEGMENT, CUSTOMER.REGION, CUSTOMER.SHIP_TO),
PRODUCT WHERE PRODUCT.SHOULD_PRECOMPUTE = 1]]>
</PrecomputeCondition>If you recreate the cube from the XML with this condition, then the PVSET valueset you discovered should contain all dimension members for which the attribute value is 1.   This gives you complete control over what is precomputed.   Note that AWM doesn't support this form of condition, so it won't show up if you go to the Precompute tab, but it is valid for the server. The PL/SQL below will modify the PrecomputeCondition (for the cube named MYCUBE) without going through AWM.
begin
dbms_cube.import_xml(q'!
<Metadata
Version="1.3"
MinimumDatabaseVersion="11.2.0.2">
<Cube Name="MY_CUBE">
    <Organization>
      <AWCubeOrganization>
        <PrecomputeCondition>
          <![CDATA[
           "TIME" LEVELS ("TIME"."MONTH","TIME".CALENDAR_QUARTER, "TIME".CALENDAR_YEAR),
           CHANNEL LEVELS (CHANNEL.TOTAL_CHANNEL,CHANNEL.CHANNEL),
           CUSTOMER LEVELS (CUSTOMER.MARKET_SEGMENT,CUSTOMER.REGION,CUSTOMER.SHIP_TO),
           PRODUCT WHERE PRODUCT.SHOULD_PRECOMPUTE = 1]]>
        </PrecomputeCondition>
      </AWCubeOrganization>
    </Organization>
</Cube>
</Metadata>
end;
/

Filter on a dimension with a fact that don't depend on that dimension

Hi experts,
I have a logical column (fact1) that it is set as total for logical level for one dimension (dim1). (it doesn't depend on that dimension but i have to put it in a report next to that dimension).
If I don't filter the report the result is:
dim1 dim2 fact1
a x 1
a y 2
b z 3
b w 4
If I filter the dim1 on 'a' then the result is:
dim1 dim2 fact1
a x 1
a y 2
null z 3
null w 4
I don't want to see those 2 lines with null...
How can I achive that...
Thank you in advance...
Best regards
Nicolae Ancuta

Thank you for your post:
Here are the details:
Phisical layer:
Stores
Products
Suppliers
Stock joins with products and with stores
Purchase Price joins products, stores and suppliers
Sale price joins products and stores
Avg sale per day joins products and stores
BMM:
Dim Stores table
Dim Products table
Dim Suppliers table
Fact stock table with source Stock inner join Sale price inner join Avg sale per day
Fact stock table joins on logical level with Dim Stores and Dim Products
The source table stock has the following mapping on logical level:
detail on - Dim Stores table
detail on - Dim Products table
total on - Dim Suppliers table
Fact stock table has the following fact columns:
Fact stock column total on Dim Suppliers table
Fact sale price column total on Dim Suppliers table
Fact Avg sale per day column total on Dim Suppliers table
Fact Purchase Price table with source Purchase Price
Fact Purchase Price table joins on logical level with Dim Stores and Dim Products and Dim Suppliers
The source table Purchase Price has the following mapping on logical level:
detail on - Dim Stores table
detail on - Dim Products table
detail on - Dim Suppliers table
Fact Purchase Price table has one fact columns: Fact Purchase Price column
The report on Presentation:
Store | Product name | Supplier name | Fact purchase price column | Fact stock column | Fact sale price column | Fact Avg sale per day column
Store 1 | IBM server | IBM corporation | 1000$ | 10 pcs | 2000$ | 0.5 pcs
Store 1 | IBM laptop | IBM corporation | 500$ | 10 pcs | 1000$ | 0.5 pcs
Store 1 | DELL server | DELL corporation | 1500$ | 5 pcs | 2800$ | 0.2 pcs
Store 1 | DELL laptop | DELL corporation | 700$ | 5 pcs | 1200$ | 0.2 pcs
if I filter Supplier name with IBM corporation my results are:
Store 1 | IBM server | IBM corporation | 1000$ | 10 pcs | 2000$ | 0.5 pcs
Store 1 | IBM laptop | IBM corporation | 500$ | 10 pcs | 1000$ | 0.5 pcs
Store 1 | DELL server | null | null | 5 pcs | 2800$ | 0.2 pcs
Store 1 | DELL laptop | null | null | 5 pcs | 1200$ | 0.2 pcs
Regards
Nicolae Ancuta

Cascade filter doesn't seem to reinitialize

I'm using a cascade filter with butterworth coefficients in the attached vi. There are some values for input I can give the filter such that it causes the output to be an array of NaN. This is fine, except that when I give the filter the exact input that has already yielded a good result, it still outputs NaN. If I change the boolean input init/cont to false, run it once, then change it to true, the filter works fine. Is the data somehow getting left over in the filter? Or is there something I can do to stop this from happening? Thanks.
John
Solved!
Go to Solution.
Attachments:
DualPhase_LockIn.vi ‏22 KB

Hi John,
Turns out I also need to read the LabVIEW Help better. If you look at (http://zone.ni.com/reference/en-XX/help/371361F-01/lvanls/iir_cascade_filter/), this constant works opposite than what we want. Essentially, it initializes the VI when set to false and uses previous data when set to true. What you can do in your VI to make sure each run starts with fresh data is to use the First Call? function and invert it. When the First Call? is called at the beginning of execution, it outputs True; False is output for all other iterations. If we invert this, you get False (Initialize in this case) on the first iteration of the Cascade filter and True for every other call of the VI.

MDX : How to use filter on a dimension without showing it OR merge 2 attributes of the same dimension?

Hello everyone,
I'm kind of newbie with MDX language, I'm trying to get filtered measures from a cube with a specific layout.
Here is my exemple :
I want to get all the "order numbers" where the "invoice" cost is more than 300. Each invoice is linked with an ID, one order number can have several invoices.
So what I do is a select request with 2 dimensions : "order num" and "id_invoice"
Here is the result : I get all the cost sorted by "order num" and I can easealy identify those who are above 300
order num
Id_invoice
cost
Total
Total2
15450307
10032013094498
1671
1737
ok
15450307
10032013099835
66
1737
KO
15450596
10032013094078
8524
8524
ok
15451214
10032013094119
228
4348
KO
15451214
10032013097069
978
4348
ok
15451214
10032013097071
3142
4348
ok
15542434
10032013101736
142
142
KO
15550064
10032013097016
1261
1261
ok
15550346
10032013097586
2252
2252
ok
15550353
10032013098446
2617
2617
ok
What i need : In "cost", I need to have the sum of each cost which are above 300 in one line (without showing the invoice id)
order num
cost
Total
Total2
15450307
1671
1737
ok
15450596
8524
8524
ok
15451214
4120
4348
ok
15550064
1261
1261
ok
15550346
2252
2252
ok
15550353
2617
2617
ok
How is it possible to use a filter on a dimension without showing it or merge 2 attributes of the same dimension ?
The problem is that if I show "id" column, the filter doesn't work properly : it compares 300 with the total cost of the whole order and does not compare with each invoice (which could be under 300 ! - see "KO" in Total2 column)
Here is the request I use to generate this result :
with
member [Measures].[Total]
as (
[TP].[ID INVOICE].[All],
[TP].[ORDER NUM].currentmember,
[Measures].[COST]
member [Measures].[Total2]
as (
IIF([Measures].[COST] >
300, "ok",
"KO")
select
{[Measures].[COST],[Measures].[Total],[Measures].[Total2]}
on
columns,
Filter
             [TP].[ORDER NUM].[ORDER
NUM].members *
             [TP].[ID INVOICE].[ID
INVOICE].members
             ,[Measures].[COST]
> 0
on
rows
FROM [MYCUBE]
Thank you for your help !
Kind regards,
Olivier.

Hi Olivier,
hereunder a similar query against the AW. For all order numbers, we sum all lines above 30 without diplaying the line numbers:
WITH
MEMBER [Measures].[lines above 30] AS
IIF(
[Measures].[Internet Sales Amount] > 30,
[Measures].[Internet Sales Amount],
null
MEMBER [Measures].[aggr lines above 30] AS
SUM(
exists(
[Internet Sales Order Details].[Sales Order Line].[Sales Order Line],
[Internet Sales Order Details].[Sales Order Number].currentmember
[Measures].[lines above 30]
SELECT
[Measures].[Internet Sales Amount],
[Measures].[aggr lines above 30]
} ON 0,
[Internet Sales Order Details].[Sales Order Number].[Sales Order Number] ON 1
FROM ( SELECT
[Product].[Category].[All Products] ON 0,
[Internet Sales Order Details].[Sales Order Number].&[SO51177]:
[Internet Sales Order Details].[Sales Order Number].&[SO51178]
} ON 1
FROM [Adventure Works] )
I reduced the cube with a subselect.
Philip,

Primary Key and Chronological Key for Time Dimension

Could someone please shed light whether it's better to use normal PK or Calendar Date (which is a chronological key on the most detailed level). I tried both and haven't noticed any difference. But my understanding is that I still need to have the PK key on the most detailed level in time hierarchy.
Thanks

Matt,
the PK of the Time Dimension is a regular sequence numeric key. I have it defined as a logical PK for the dimension. However, I guess I was talking about hierarchy key. I tested it either way and the query time is the same. So I guess it doesn't really matter.
--------Year
Year (chronological key and primary key)
------------------Quarter
Quarter (key)
Quarter+Year (primary key, chronological key)
---------------------------------Day Detail
PK (key)
Date (primary key , chronological key)
I hope it makes sense.

Limit rows for Degenerate dimension in SSAS 2012 Standard Edition

Hi Experts,
I have a fact table consisting of 4 measures and 6 descriptive fields. The Geography and Time dimension are separately palaced. I want to build SSAS cube on this schema.
I have created the dimension from fact table itself for the descriptive fields. The scenario is that we are not loading complete fact table in SSAS. Only recent 2 years out of 10 years is brought into SSAS. But the degenerate dimension is procssesing all
the rows from fact.
I want to limit the dimension processing to only recent two years of data. How will I be able to achieve this?
Please help me out of this.
Thanks,
RuchikaG

Use a view as the source for dimension. Inside that write logic to filter data from the fact for last two years.
ie like
WHERE dateField >= dateadd(yy,datediff,yy,0,getdate())-1,0)
AND dateField < dateadd(yy,datediff,yy,0,getdate()) + 1 ,0)
Then take distinct attributes from it for the dimension
Please Mark This As Answer if it solved your issue
Please Vote This As Helpful if it helps to solve your issue
Visakh
My Wiki User Page
My MSDN Page
My Personal Blog
My Facebook Page

Entering value for parent dimension

Hello
Once the dimension are loaded from OGL into Hyperion along with hierarchy (parent with child relation) is there a way to enter values (financial figures) for parent dimension ONLY into the dataform manually?
If yes, can you please suggest a way how to go about entering values....

You would probably need to use a target type version to enter at upper levels
Cheers
John
http://john-goodwin.blogspot.com/

Alternate hierarchy dimension build with load rules

Hi,
I am trying to build a alternate hierarchy dimension using load rules (I feel using EIS is easier to build ASO and multiple hierarchies but we r trying to get rid of EIS member loads). Looking at different posts my understanding is it has to be done using two different rules file first to build the stored hierarchies then use parent-child reference to build the shared hierarchy. Below is the link for the picture how I am trying to build the dimension. I've tried different possibilities but no luck, I am having a tough time to come up with a way to do this using rules file.
http://s24.postimg.org/u5hl4eyqt/basis.jpg

If your file is in a parent/child order, then you can build easily in one rule. make sure the primary (stored Hierarchy) gets built first followed by the alternate hierarchies. I suggest adding a SortID column to the source so you can sort the file to get it into the right order. You can also build the alternate hierarchies using level or generation builds in one load rule, but especially if you have a lot of alternate hierarchies or members, then the parent/child method is easiest.
Read in the DBAG about parent/child files and it will give you more info

Cascade Filter for Hierarchy dimension

Similar Messages

Maybe you are looking for