Hello! I have a few questions about dimension anomalies. I created DA tests for a couple of models with the following config:

- elementary.dimension_anomalies:

dimensions:

- <dimension_1>

- <dimension_2>

- <dimension_3>

min_training_set_size: 1

tags: ["elementary"]

backfill_days: 1

days_back: 1

config:

severity: warn

1. It's been more than 3 days but it 's still showing 'Not enough data to calculate anomaly score'. I remember it wasn't the case with volume anomalies - there it was showing the graph after one day with the same config. Do these tests work differently? 2. I need to create test(s) that would group by 3 columns SEPARATELY. This implies creating 3 different blocks of code, right? And the attached config makes elementary group by 3 columns at once? 3. When I artificially change row number (using LIMIT), I get a warning from elementary, but when I want to see some details, I see what's on the attached screenshot. Why aren't those fileds filled?

Hi @Denys, 1. Do you have a timestamp column configured on the table? I don't see it in the test config, I recommend adding one if possible. if you do that you'll see the results after one run, assuming there's enough data in the table. In addition, you set up

min_training_set_size = 1

but you have 0 training set because your days_back equals your backfill_days (so your detection period is 1 and your training period is in fact 0). If you set up days_back=2 that'll match. but I will stress this is not a recommended setup, and using more days with a timestamp column is better and also has immediate results. 2. Exactly right 3. There's a bug in the latest version that causes the fields to be unpopulated. We're working on a fix, in the meantime you can see the details in the result description section

Thank you for a quick response!

hmm interesting 🤔 can you try to run the test again with the flag:

--vars '{"force_metrics_backfill": true}'

?

The same. Says there's not enough data. Here's test runs details.

what do the test results show?

image.png

@Denys just to be sure, do you have anything configured in the model level or in the dbt_project.yml that is related to this test?

Nope. The only thing related to Elementary in my dbt_project.yml is this:

elementary:
    +schema: "elementary"

and the table does have data I assume? Just making sure