Is there any way you can post a dummy version of your waterfall chart...I'm not 100% sure why the "exclude zero" doesn't do the trick, so would be good to see your setup. There are also (dynamic) ways of limiting an Axis, so happy to take a look (always love a good Axis challenge!)
Here's a video from Joe on how you can create an anonymised workbook (simply)
Thanks for the quick reply Simon - seems like you're expert in this area!
I've attached for an anonymized version of the tableau file as you requested. Please note, the data values are slightly different to the one in the screenshot of the original post, however the problem on non-zero axis still persists.
As far as I can tell, if I remove the secondary measure for the gant box sizes (see second tab "_wo_gant") the axis appears to scale correctly, but I lose the waterfall effect.
Waterfall Demo_external.twbx 218.5 KB
Thanks for the workbook...and it does make sense now. The Gantt size is drawing the gantt marks back to the axis, which is why it always goes to zero.
I've has a bit of a plauy, and can get part of the way there by using your
(SUM([Defaulting Balance Diff])) = LOOKUP(ZN(SUM([Defaulting Balance])), FIRST())
test, to only draw the first mark back to 10%
IF (SUM([parameter_diff])) = LOOKUP(ZN(SUM([parameter_diff])), FIRST()) THEN
SUM([Defaulting Commitment Diff])*0.1
SUM([Defaulting Commitment Diff])
but I can't affect the GT...so we get this
Close...but no cigar!!
I'll have a bit more of a think, and hopefully come up with something tomorrow (or call in the big guns...ie. Jonathan and/or Yuri!!)
Thanks for quick turnaround Simon - this is already a good start. I'm wondering whether I should create a 'final' step row in the dataset to forgo the need for a total.
If you have that freedom...that would make things much easier! We can the use the Last() on Lookup, to pick up that stage & do something similar to draw the gantt back to the point at which the first mark ends.
If you can do this, let me know and I can definitely give you a solution!
Sent from my iPhone
I do have that flexibility - thankfully - and have updated the dataset to include a 'step_final' data point. I've also made some updates to the way the diff (gant width bars) are calculated and included your suggestion of checking for first() and last() to reduce the scaling and it works like charm.
I had just had to do one extra step to ensure the synchronization of the gant height for the first and last bar which I achieved using this ugly formula. It takes the difference between the bar value and 70% of the window minimum to calculate what the gant height should be for the first and last step.
IF (FIRST()=0 OR LAST()=0) THEN (SUM([Defaulting Balance])-(WINDOW_MIN(SUM([Defaulting Balance]))*0.7))
ELSE (SUM([Defaulting Balance])-ZN(LOOKUP(SUM([Defaulting Balance]),-1)))
Thanks again for the help, really appreciate it!
Waterfall Demo_external_v2.twbx 188.1 KB
Great work Tristan,
I hadn't fully thought through the logic of the "Final Step" adjustment...but knew we'd have to look at what length we set the "Step 1" to and then adjust "final step" so they end at the same point on the axis. Although I hadn't thought through the logic, one thing I did know "I ain't gonna be pretty"!!
Some pretty advanced stuff there