I've been using the Classic version of Azure ML to teach my graduate-lvel professional MBA/EMBA students. It was fantastically easy for them to set up and run their very first experiment in 10 minutes or less. They were also able to run each step of the experiment individually, which is really helpful for succeeding steps have access to the results of the previous steps and don't require the student to type in names of variables.
The new version of Azure ML is a disaster. Tech savvy as I am (with 35 years of experience as a software engineer and executive), it was monstrously difficult to set it up. After final set up, I couldn't get even one experiment to run. Even starting up the compute instance took several minutes.
I kept getting this error:
"Failed to execute command group with error An unexpected error occurred while executing command due to: Failed to pull Docker image mcr.microsoft.com/azureml/runtime/cap/lifecycler/installed:eastus2-stable
due to: None"
At this point, I'm throwing in the towel. Even if this is solvable, I'm sure there will be hundreds of variations on this and other types of errors. All this has nothing to do with learning ML. My students will get stuck in the weeds of Cloud configuration, instead of actually learning ML.
Retiring the Classic version is a terrible decision by Microsoft, not well thought out at all. You've basically lost your best influencer advocates - the MBA/EMBA students, who would have influenced buying decisions in their companies.
If you can, please retain the Classic free version for students with 'edu' emails. Please contact me for further feedback or to discuss.