You could try reference lines, they would work for some scenarios. A dual axis *might* work, pulling out the top value into another field - although I've not tried this with a gantt and it may be a non-starter, I'm just throwing an idea out there.
If you provide a workbook with sample data someone using the forum will no doubt put something together.
Reference lines are too inflexible. Dual axis is also problematic, since I want to run a R script afterwards on each row. With a dual axis chart I probably wouldn't be able to define the required partitioning.
Nevertheless, I attached a workbook. Basicaly I want to move the rows with the red gantt bars into each other row.
demonstration.twbx 1.2 MB
This is very tricky, I am stumped and sadly don't have the time to try and work it out. You basically want a date dimension to ignore any other dimensions trying to slice it so some sort of table calculation may allow you to do that. Putting this value, which is against one value in a dimension, against each other value in a dimension (category in this case) is tricky.
I also considered LoD calcs, thinking FIXED would do this, but not sure if this would work as it's a date and you don't want to aggregate the date, you just want to return the date, so also not sure how this would work.
Hopefully someone else out there can help with this, I'm interested in the solution.
Well, I know how to solve it programmatically and with TDEs, changing the underlying data structure, however, then I'd loose all flexibility of parameters (won't be able to change the view based on them).
I took a quick look at this. The challenge here is that in order to draw the marks there either need to be extra measures (for each date) or extra dimensional values to generate the marks. I had a similar “what about LODs?” idea as Andrew but I can’t come up with something that works because the data is sparse, not every Furniture date exists in every Technology or Office Supplies date.
So we need extra dimensional values. Therefore my suggestion would be to pad out the data, here are two ideas:
1) Making a data source that has a record for every combination combination of ID & red dates & green dates, with a red/green dimension to identify which was which and that dimension would go on the Color Shelf.
2) Make a data source that has every combination of possible date & ID and use that as the primary source in a data blend with two secondary sources, one gets the red marks and one gets the green marks. I don’t know how many marks you’re wanting to display at once, if the number gets into the thousands or tens of thousands then the view might not have the performance that you want.