Forum Discussion
Filter Function Help
Might need a small pic of your model showing the relationships. Are the construction/transaction tables related by ID, with construction id a unique value?
transaction table:
| ID | transaction date | time segment | room number | visitors | revenue |
| 4652 | 1/1/2010 | Morning | 1 | 1 | $ 7.50 |
| 7854 | 1/2/2010 | Afternoon | 2 | 5 | $ 37.50 |
| 9635 | 1/3/2010 | Before Noon | 3 | 4 | $ 30.00 |
| 7845 | 1/4/2010 | Morning | 4 | 6 | $ 45.00 |
construction table
| ID | Construction Start | Construction End |
| 4652 | 1/1/2010 | 3/15/2010 |
| 7854 | 8/5/2011 | 12/7/2011 |
| 9635 | 9/7/2011 | 1/1/2012 |
Date key table
| Date | Year | Month | Week | Day |
| 1/1/2010 | 2010 | 1 | 1 | 5 |
| 1/2/2010 | 2010 | 1 | 1 | 6 |
| 1/3/2010 | 2010 | 1 | 1 | 7 |
Tables are linked together by unique Building ID and I also linked everything to the Date table (in hopes of being able to manipulate views/cuts of data later).
Ya, you will have an easy time w/ filtering transactions by date, and filtering transactions by construction... but you have an issue w/ your calendar table relating to construction table... since there are 2 dates.
If you make a relationship between your calendar table and construction on Construct Start... then filtering to June 2010 is going to show ONLY construction that STARTED in June, which is... maybe not what you want?
Let's talk about a specific measure to make this easier? Like... you want... revenue year to date, broken out by construction?
I could make the construction 2 different tables. That would be an easy enough thing to do. What I'm trying to show is total revenue 12 months before construction starts, revenue 12 months after construction ends...and ideally any number of quarters, months etc. So the apples to apples comparison wouldn't be in individual months/years, but instead be based on number of days before or after construction started.
"1 year after renovations were completed, location 1 had x revenue, while location 2 had y revenue after it's remodel (that ended at a completely different time)."
"location 1 construction ended 1/1/2010, in 6 months it made $x. In the same 6 months before construction started location 1 made $y."....basically looking at the effectiveness of construction....
Thank you so much for the help!
I can imagine various ways of creating multiple relationships between the same tables, then using USERELATIONSHIP to "activate" the relationship you need... but I dunno, I can't decide if that is a good or bad idea. Maybe somebody else has an opinion :)
Let's pretend both date columns in your construction table don't have any relationships.
Revenue - Year After Construction Ends := CALCULATE( SUM(Transactions[Revenue]), FILTER(ALL(Transactions[Date]), Transactions[Id]), Transactions[Date] > MAX(Construction[EndDate]) && Transactions[Date] <= MAX(Construction[EndDate]) + 365 ) )Give me sum of revenue, but only for those transactions with a date starting at end of construction, and ending after 1 year...
That ... works in a weird way when looking at multiple constructions... (uses the max ending date of any of the constructions), so you probably want to wrap that in a SUMX to iterate over all the constructions... one at a time, adding their revenue together.
Revenue - Year After Construction Ends := SUMX(Construction, CALCULATE( SUM(Transactions[Revenue]), FILTER(ALL(Transactions[Date]), Transactions[Id]), Transactions[Date] > MAX(Construction[EndDate]) && Transactions[Date] <= MAX(Construction[EndDate]) + 365 ) ) )
I could imagine that... doing something... :)