What you're trying to achieve is not possible with data blending.
It is mainly because of your event datasource looks like an accumulative snapshot
(each row is a unique event id, and it has several "status" date fields, such as start / end date).
But for the question you've asked another data structure is best suited, namely a periodic snapshot.
There would be several rows for each unique event id -- as many as days from start to end date --
and each row has a unique event id / date combination. Then one could blend holidays on a date field.
Having datasources like yours, a couple suggestions could be made:
1) If possible, put your holidays table into the same database scheme
(of the underlying DBMS instance) as where your event table resides.
Then do a non-equi join on dates, and prepare an extract from that.
2) If you're able to read-only from an underlying DBMS,
then you could use Custom SQL (a UNION ALL Select)
to Pivot the event table -- by aligning both original fields
(Start Date and End Date) into a pair of new fields
(say, Pivot field names and Pivot field values).
Having a datasource like this, you'd be able to blend
your Holidays (Excel) datasource on a Date field.
Please look at the attached as an example of the latter.
Hope it could help.
PS Tableau 10 Cross-Datasource Joins couldn't help with the 1st approach,
because of non-equi joins not possible across DS, hence 2nd approach only --
but with a 'regular' left join the subsequent calculations could be built
a bit simpler than with a 'conventional' blend.
Holidays_Count_YF.twbx 20.5 KB