question

achrafdridi-4443 avatar image
0 Votes"
achrafdridi-4443 asked GiftA-MSFT edited

AutoMLException: Message: Could not find a model with valid score for metric 'accuracy'. Please ensure that at least one run was successfully completed with a valid score for the given metric.

Hi, I am trying to make a classification using automl service , I've chosen KNN as model to train and for the primary metric i used accuracy . Based on the automl documentation the metric accuracy can be used with this type of classification task, but I get the error : AutoMLException: Message: Could not find a model with valid score for metric 'accuracy'. Please ensure that at least one run was successfully completed with a valid score for the given metric. when I checked azure ml studio I find this error of the run is that means that the dataset that i am using can be trained on KNN model ? ![197051-image.png][1] [1]: /answers/storage/attachments/197051-image.png

azure-machine-learning
5 |1600 characters needed characters left characters exceeded
Toggle Comment visibility. Current Visibility: Visible to all users

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

1 Answer

GiftA-MSFT avatar image
0 Votes"
GiftA-MSFT answered GiftA-MSFT edited

Hi, thanks for reaching out. I can confirm that 'accuracy' is a valid primary metric for classification tasks. It's possible that your dataset isn't compatible due to high dimensionality, missing or sparse data. If you click on the error message details, there might be some more information that can help understand what caused the error. Review Data guardrails to identify the issues with your dataset. Also, try to allow additional models to see if it helps.


5 |1600 characters needed characters left characters exceeded
Toggle Comment visibility. Current Visibility: Visible to all users

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.