Simple Beast Mode Failing on Federated Dataset

LCash

I've got a federated dataset pulling a redshift table and I want to perform some simple calculations like SUM(column1) / COUNT(column2).

However, this calculation is resolving to 0 consistently despite the correct value being about 0.2.

If I write a beastmode for SUM(column1) I get the correct value. And if I do the same for Count(column2) I also get the correct value - just not when I divide them.

My hypothesis was that this was some kind of timeout issue with our database connection settings, like this one: https://dojo.domo.com/discussion/53003/beast-mode-processing-limitations

But 1) I also get some pretty funny values if I write beastmodes like 5/4 (this resolves to 1.00000!) or 5/6 (this resolves to 0.0000).

AND 2) I've limited the dataset to just 1000 rows

So I'm wondering whether it's not something else (a weird rounding issue?)

Has anyone experienced anything similar or could point me in the direction of what I could do to resolve?

GrantSmith

Hi @LCash

Just curious if you multiply your numerator and denominator by 1.0 to force them both to be decimals on the off chance that they were being treated as integers and doing integer division instead of decimal / float division.

(SUM(column1) * 1.0) / (COUNT(column2) * 1.0)

amehdad

Hi @LCash, alongside this thread and awaiting the advice of the technical adept users, reach out to Support [support.domo.com] to have them look at it as well (and escalate to Dev if need be). 

MarkSnodgrass

In your card, how do you have the field formatted? If you are leaving it as default it may be rounding it to the nearest whole number. Make sure you click the pencil icon when hovering over the field and choose Format and Display As Number and choose the appropriate decimal places.

LCash

Thanks both! I will raise a support ticket. To confirm @MarkSnodgrass decimal precision is set to 6 but still some weird rounding (or something) going on. Have tried ROUND((5/4),6) as well, same result.

The same beastmodes on a non-federated dataset work fine - and like I say other beastmodes on the federated set seem ok too, so I'm at a loss. Hopefully support can point me in the right direction.

MichelleH

@LCash I second @GrantSmith's suggestion. I've run into a similar situation in some of my SQL dataflows where the result was rounding to the nearest integer because the numerator and/or denominator were integer datatypes. Forcing both to a decimal should do the trick.

LCash

Thank you @GrantSmith and @MichelleH! That's worked.