Weird behavior of (left) join recipe with post-join computed columns losing records

Options
Antal
Antal Dataiku DSS Core Designer, Dataiku DSS ML Practitioner, Neuron, Dataiku DSS Adv Designer, Registered, Neuron 2023 Posts: 85 Neuron

I'm experiencing something unexpected with the join recipe using a left join and post-join computer columns.

I'm joining 2 datasets on a single column and then computing 4 additional columns after the join. I check the number of records before and after the join.

With 3 of my computed columns everything's fine and the number of records stays the same after the join (about 260k). But when I add the 4th column consisting of a series of case when statements on a single column, I end up with only 80k records in the output dataset.

None of the computer columns use the column used for the join.

Furthermore, if I use VIEW QUERY on the OUTPUT tab and run that query in my database, I do get the normal 260k records. So it's not a problem with the built query.

I'm using in database SQL processing in a Teradata database.

Any ideas what could be causing this?


Operating system used: Windows

Answers

  • Sarina
    Sarina Dataiker, Dataiku DSS Core Designer, Dataiku DSS Adv Designer Posts: 315 Dataiker
    Options

    Hi @Antal
    ,

    In the future (or if this is still an issue!) please feel free to open a ticket with the full job diagnostic along with screenshots showing the unexpected output so that we can help troubleshoot.

    Thanks,
    Sarina

Setup Info
    Tags
      Help me…