<#C02CTC89LAX|> <@U02C1PLFA49> <@U052AVD565T> <@U...
# community-support
v
#C02CTC89LAX @Or Avidov @Noy Arie @Noa Kurman Hello Team, I am trying to add dimension anomaly, in my configuration .yml file I am adding column with lowest cardinality, however, I am unable to see the graph for expected vs actual.
Copy code
- name: table
    tests:
      - elementary.dimension_anomalies:
          dimensions:
            - COLUMN_1
          where_expression: "COLUMN_2 = RED"
          # timestamp_column: "DATE(ORDERDATE)"
          # time_bucket:
          #   period: day
          #   count: 1
above is my configuration Please help This kind of graph is missing
e
Hi there @vinit, We haven't yet implemented visualization for dimension anomalies unfortunately 😞 This is more complex since we need to show a graph for every dimension separately. We are working on this and will release this soon!
v
I understood, But I am not getting anything on dashboard
@Ella Katz .
Thanks for the information @Ella Katz Here I am scratching my head since last 10 days
e
that doesn't explain the "not enough data" message though... I see in your test description the timestamp column is commented out - why is that?
v
is it compulsory to specify timestamp column ? docs says it will simply calculate total no of rows if timestamp column not mentioned @Ella Katz
e
ah, yes. but in that case you will only have enough training data after the test runs for as long as your "days_back" is configured, which in this case is the default of 14 days
in addition, I see that we have an open issue that happens when you do set up a timestamp column but not in the test level and you may have ran into that as well; https://github.com/elementary-data/elementary/issues/1170
v
yes exactly
thanks @Ella Katz I will monitor this ticket, is there any ETA for this ?
c
Hi @Ella Katz have you folks got around to doing the Viz for this? I think any viz that shows its running for past days for useful, either a bunch of line graphs overlayed eachother or maybe a stacked bar chart would be even better