No grand total if Time Intelligence Calculation is drag on row

Hi,

I've deployed a simple cube with a time dimension on SQL 2005. I added the time intelligence which generated some calculations like Day over Day difference, Month over Month difference. If I drag it to row on a pivot table, the grand total of the measures cannot be shown.

Is it because the calculated member is not aggregatable? Is there any ways to show the grand total as well as those calculation on the pivot table?

Million Thanks.

KJ

Kelvin,

This is the intended behaviour. The calculations are written in such a way as to purposely place "NA" in cells where the calculated value would not make sense. For example, if you think of a YTD calculation in the context of all years, it doesn't really make sense; likewise for a year over year calculation.

-rob

No data in Data Mining cube.

Hi,

I've created a Cluster data mining model and seems to correctly return data. However I've created a data mining dimension and cube but when I go to query the cube it doesn't return any data if I select any members on the data mining dimension. Any suggestions to where I can look to resolve this?

Thanks

Sanjay

Hi,

I found out how to see the data in the cube. I was using the data mining dimension in relation with the dimension it was based on. So now I can see data for the years that I've trained my model on. Is there anyway to make the cube predict the values for the other years in the cube with data?

Thanks

Sanjay

|||

Are you saying that for example you've created a cluster model on a set of items, and you want to see the cluster values for items not in the set? And you want the slicing behavior to work against those values as well?

If this is the case, unfortunately, that's not the way it works. The cluster dimension is only on the cases for which it was trained. You could apply the clustering to the data outside the cube, write the results to a table and process it as a regular dimension, though.

|||

Hi,

Ideally yes. For example if I generate a cluster model and train it with a couple of years data, potentially I would then like to use the data mining dimension and cube to predict or classify the current years data. I will look to use your suggestion.

Thanks

Sanjay