Azure Machine Learning is the home for Data Scientists to run experiments. But the language and constructs used in Microsoft Azure ML will be confusing in the first go. This article helps to clear some fundamentals.
The below words are confusing, and the UI is not that great.
- Pipeline
- You can create a pipeline object and execute in an experiment.
- Pipeline Job
- Experiment: Pipeline jobs are grouped into experiments to organize job history. You can set the experiment for every pipeline job.
- Pipeline Endpoint
- Can have multiple Pipeline but only one could be the default.
- Published Pipeline
Running a script - Experiment Job Type: Command
from azureml.core import Workspace, Experiment, Environment, ScriptRunConfig
ws = Workspace.from_config()
experiment = Experiment(workspace=ws, name='day1-experiment-try1')
config = ScriptRunConfig(source_directory='./', script='Hello.py', compute_target='gurupc')
run = experiment.submit(config)
run.log('mymetric', 1)
run.wait_for_completion(show_output=True, wait_post_processing=True)
aml_url = run.get_portal_url()
print(aml_url)
Running a Pipeline - Experiment Job Type: Pipeline
https://github.com/guvijaya/TipsAndTricks/blob/master/Azure%20Machine%20Learning/AML%20Pipelines%20Intro.py
