Data Drift Detection Issue after DSS v12 Update

[Suhail]

Hey Dataiku Community,

I'm facing an issue with the data drift detection feature in the Evaluate recipe after updating to DSS v12. Prior to the update, this feature was working perfectly fine. However, after the update, I'm encountering an error that's preventing the Evaluate recipe from running successfully.

Error Message:

Oops: an unexpected error occurred

Error in Python process: <class 'dataiku.doctor.exception.DriftException'>: Failed to compute psi for column sum_of_values : `bins` must increase monotonically, when an array

I've tried a few things to troubleshoot the issue, such as checking the recipe configuration, verifying data inputs, but nothing seems to resolve the problem.

Has anyone else encountered a similar issue with the data drift detection feature after updating to DSS v12? If so, could you please share your experiences and any potential solutions you've come across?

Any insights, suggestions, or workarounds would be greatly appreciated.

Thank you in advance for your support and input!

---

Operating system used: MacOs

[gambinoga]

Hello Suhail,

This looks like it could use the precise data and logs to be able to fully replicate and diagnose. If you are able to privately share those with Dataiku for troubleshooting, I suggest opening a support ticket (by writing to support@dataiku.com). Meanwhile, how many different values you have in your column 'sum_of_values' ?

Also, note that you can choose to ignore those errors that happen on univariate drift computation by going in the Advanced section of your Evaluate Recipe configuration, and uncheck the "Fails on warnings" of the Drift section.

Gregoire

[Suhail]

Hi @gambinoga,

thanks for the response 

sum_of_values is a numerical column with values ranging from -0.12 to 12

unchecking the "Fails on warnings" is throwing the following error

Error in Python process: <class 'Exception'>: Error : Streaming: push-data call failed: b"Session 9CK8kRaZI5 doesn't exist"