Tabular Model gives different results than SQL Query

[jbirrer]

I am having trouble with some records and how they are aggregating inside our Tabular Model compared to our Data Warehouse.

I have a custom query built that pulls a record number, the order of the records, and a count associated with those records.

When i run the source query (same that is in the tabular model) against the DWH directly, i get this result, which is correct:

When I pull the same fields directly from the Tabular model (from a table built using the query yielding the above results), I get this result, which is incorrect:

I get the same result browsing the model directly in SSMS and if i import the data to Power BI. I have tried disconnecting all relationships from the table in the model and it does the same thing. I have traced the DAX query and executed it against the SSAS DB directly and I get the same result (incorrect). I should be seeing a count of zero on OfferOrder 1-4, but for some reason, the Tabular Model is calculating a count of 1 on the 4th offer. I don't really know what else to do at this point. Any ideas?

[v-lid-msft]

Hi @jbirrer ,

Could you please share the formula you are using to generate "OfferTransportCount" if it does not contain any confidential information? Is there any record in your datatable that OfferOrder=4? Please don't have any Confidential Information or Real data in your reply.

Best regards,

[jbirrer]

@v-lid-msft 

Here is the raw data for this "Call."  The only thing that stands out to me is that Offer #5 happened the next day (which is a valid use case).  The OfferTransportCount field is just a whole number value.  If i just import this data set to Power BI, the data looks fine, but for some reason, when i pull the data from our Model, it gives a TransportCount to Offer #4.  I am using the exact same query when pulling the data directly from the database and as the source query for the tabular model table.  Does this help at all?

CallNumber	OfferDate	OfferTime	OfferOrder	OfferTransportCount
19-11988A	2019-04-30	18:07:21.0000000	1	0
19-11988A	2019-04-30	18:07:38.0000000	2	0
19-11988A	2019-04-30	18:11:43.0000000	3	0
19-11988A	2019-04-30	18:11:53.0000000	4	0
19-11988A	2019-05-01	00:42:02.0000000	5	1

[jbirrer · Accepted solution]

This issue is resolved now.  I have partitions on some tables in our Model and the date I was partitioning on was the wrong date, so the records were being counted in both partitions instead of having the entire record sitting in only 1 partition.