Pre-built ML solution which seemingly worked fine has suddenly stopped? - Similar Resolved Incidents

[AmenaA]

Similar Resolved Incidents ML solution: somehow says "Unable to Predict" now, while it was working fine earlier.

What could possibly be the cause of it when no changes were made to the solution? It was only ever retrained. 

Appreciate any thoughts / comments on this. Thanks In Advance.

[AmenaA · Accepted solution]

This is shown when data it's been trained is not sufficient enough to predict.

[Lener Pacania1]

Did you try lowering the threshold and testing again.  The threshold sometimes does not save when retrained.

[AmenaA]

Hello @Lener Pacania1 ,

Thank you for your reply on this. 

It was using data created on past 6 months. Starting this Month, there had been no incident's created on past 6 months over this instance causing it to not be able to predict. Extending the filter condition to cover more than 6 Months fixed it! 🙂 

[Lener Pacania1]

Good to hear AmenaA, adjusting the filter and checking the record count and retraining was the next step if the threshold didn't work.  Glad you figured it out.