How can a reoccurring problem in the pipeline be fixed?

Question

I am working on a pipeline that consists of multiple pill models. Specifically, the pills that are the most helpful for me are the Train Model, Evaluate Model, and any of the Regression Models. After building the pipeline and then configure and submitting it, I run into the same problem every time I complete the pipeline. To answers my questions I need to access the explanation and metrics sections within the pills model. In the beginning I can see the explanation (preview) and metrics; however, after a while when I go back to take another look the sections are unavailable. This is very frustrating, because I cannot answers the questions correctly in my assignments and I cannot move on to the next question.

To resolved the issue I have done the following below:

• Ensure that the pipeline has been run successfully (validate), and all steps are complete.
• After running, right-click on the pill (instead of double-clicking) and choose “Visualize” or “Examine” to see if the explanation appears (when i right-click those option don't appear).
• If the issue persists, restart Azure MLS and reload the project, as this sometimes clears temporary glitches.
• Mark Keith addressed the issue is his YouTube titled "Azure Machine Learning Designer Model Interpretation from Train Model Explanation". His solution is to:
  • go back to designer
  • retrain the pipeline ( you have to change one thing)
  • configure and submit
  

I've tried everything I can think of to resolve the issue. Nothing seems to work. Has anyone had this particular problem with their pipeline? How did you resolve the issue?

Answer 1

Hello Ebony Johnson,

Welcome to Microsoft Q&A! Thanks for posting the question.

As I understand that you are encountering an issue with the Azure Machine Learning Designer where the explanations and metrics from the Train Model steps become inaccessible after a pipeline run, Here are some additional troubleshooting steps:

• Try to recreate the pipeline from scratch, this can help eliminate any underlying issues caused by previous configurations.
• Additionally, making small changes to your pipeline, such as tweaking hyperparameters or adding simple steps, can often trigger a refresh necessary to restore the presentations. Once you’ve made these adjustments, retrain the model by configuring and submitting the pipeline again. This approach has been effective for many users in resolving similar issues, as it ensures that all components are properly regenerated and updated.

I hope you understand, do let us know if you any further queries.

Thank you!