Why is the overall total sum amount different from the sum amount per debit credit?

[novalcia]

I'm using the Summarize tool to just sum the amount but I got contradictory results. If I just sum the total amount directly:

but if I try to total sum by debit credit:

This is from the same data file. This happened to all my data. This still happened even after I tried to rerun it.

Why is it contradictory? I don't understand what happened.

This is the Alteryx I'm using:
Version: 2022.3.1.573 Patch: 7 

Running Non-Elevated

I also tried using Version: 2023.2.1.17242 Patch: 4 but it still happened

[novalcia]

Sure

[alexnajm]

Here's a Google Drive link for all: https://drive.google.com/drive/folders/1-MOWsNmIC47UFGQRkZ4iAQsXO4Nclg96?usp=sharing

[novalcia]

the data is uploaded

[alexnajm]

I am seeing the same behavior in 2025.1.... very odd. @OllieClarke & @jrlindem & @JerryCooperCMA feel free to download it too if you want

[OllieClarke]

@novalcia 

That's really interesting. When I use your uploaded data (in 2024.2), I get different numbers.

Here are my int64 numbers (which sum to 0)

And here are my double numbers (which also sum to 0, but don't look like they should).

I would suggest using int64 for this data (which was confirmed by the autofield tool).

[OllieClarke]

@novalcia @alexnajm 

I just ran the same workflow with AMP turned off. My Double numbers are the same:

but my int64 numbers are now different

weirdly both the double and int64 E1 numbers sum to 94,560.

[alexnajm]

Crosstab yielded closer but still different numbers as well...

[OllieClarke]

@novalcia @alexnajm 

When I run the same workflow in 2025.1 (2025.1.2.142  Patch: 2) with AMP on, my double numbers continue to be the same

But my int64 numbers are good again

Running in E1 gives me a couple of field conversion errors

but the same double numbers

and the same incorrect int64 numbers as E1 in 24.2

(both summing to 94,560)

AMP being on and int64 being the data type seems to be the right move here. As for why that's not working on your machines, I'm wondering if it's a memory thing, or the fact I'm running windows on parallels on an apple silicon mac? Not really sure to be honest.

@novalcia I'd create a support ticket if I were you, this is very weird behaviour, and I don't see any reference to it in the known issues of the release notes

[OllieClarke]

@alexnajm 
I again get different numbers to you with cross-tab (int64):

and when I use cross-tab with double I get matching numbers, although different ones to my int64 summarize tool:

this is all very strange

[jrlindem]

Late to the party

Ran on 2023.1 using AMP (just for fun) and was seeing:

Similar, when using DOUBLE (bottom leg).  I get different numbers, but when aggregating without grouping, I also get "0" (expected).  I got expected results from INT64 and FixedDecimal (had to open up the precision).

Can't account for the why this issue is occurring.  Agree with @OllieClarke - Probably time to open up a ticket...  -Jay