Grouping for IQD

[NuniPR]

Hi Guys!

I have another question. I work for a Telecomm company. One customer can have more than 1 line. My model measures both customers and lines.

I have an IQD that is the source for a Model. A few weeks ago, the numbers were correct. But for the past week, the cube has been showing the correct amount of lines, but 0 customes. So, I went back to the IMR, only to discover that it had lost the grouping for the customer id. The running-count calculation of customer was still there, but it said 1 for every record, when it used to say only 1 for each group.

So, I re-grouped the report, and saved it as an IQD. Changed the datasource for the model to this new IQD and created the cube. It used to take 40 min to create the cube before this change, and ever since I re-grouped it, takes about 4 hrs.

The amount of data is still practically the same (well, data is appended every day, and still took 40 mins before), so it's not an increase of data. When I remove the group and recreate the cube, it takes 40 minutes .

Do you have any idea as to what the problem could be? Should I create a new IMR and a new IQD? Should I change the running-count calculation, and if so, how do I count the amount of customers, regardless of the lines they have? (Like I said, 1 customer can have more than 1 line.)

Thanks guys for all your support.

Nunina
Software Analyst
San Juan, PR

 

[DoubleD]

I would guess your running total is taking additional time as it may not be calculating in the database, but instead on the desktop.

I would suggest a separate IQD to count the distinct number of customers.


I am what I am based on the decisions I have made.

DoubleD

 

[NuniPR]

Hi DoubleD:

What you mean is to add a datasource IQD to my model with only the customer ID?

I can also just filter by the distinct Customer ID (maybe this is what you mean...) Can you be a little more direct, please?

Thanks a million.

Nunina

 

[DoubleD]

I mean create an additional IQD containing the appropriate dimensions but the only measure in that IQD is a Distinct Count of CustomerId.

Then add that IQD as a datasource.


Another option is to do a category count on Customer ID. If you're not sure what this is, just look up Category Count in the Transformer help.

I am what I am based on the decisions I have made.

DoubleD