https://learn.microsoft.com/en-us/azure/machine-learning/?view=azureml-api-2

Machine Learning Studio
https://ml.azure.com/home?tid=02e760b0-1b5a-4c7d-9e8f-4d421bd8cbb0

https://azuremlexamples.blob.core.windows.net/datasets/credit_card/default_of_credit_card_clients.csv

Can be downloaded from Azure Machine Learning Studio or from VS Code Desktop version.

Can be opened in VS code Desktop from Azure ML studio

Login issues in VS code or on Azure can be resolved from Azure Notebook window (banner to authenticate will pop up above ntotebook).

Folder hierarchy when loading files in notebooks refers to workspace notebook file structure (root folder is username, current folder is folder where notebook is located).

Use "Azure ML: Set Default Workspace" command to set the default workspace that exists in Azure if none is set. If it is set correctly a subscription will show workspace and resources under it.

https://learn.microsoft.com/en-us/azure/machine-learning/how-to-setup-vs-code?view=azureml-api-2

Azure managed compute instance is still required (although minimal), external can be used as added workforce.

Create a new kernel for workspace on compute instance https://stackoverflow.com/questions/65159308/how-to-use-existing-conda-environment-as-a-azureml-environment

Section "Create and Attach a DSVM as a compute target", doesnt have to be DSVM image, could create Conda environment as sections below on vanilla linux VM:
https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/training/train-on-remote-vm/train-on-remote-vm.ipynb

Initial resources for safekeeping:
https://learn.microsoft.com/en-us/answers/questions/349518/azure-machine-learning-and-attached-compute-instan
https://rukiatu.com/installing-azure-machine-learning-sdk-for-python/

Installing miniconda
https://docs.conda.io/en/main/miniconda.html#quick-command-line-install

Using miniconda
https://docs.conda.io/projects/conda/en/stable/user-guide/getting-started.html#starting-conda
conda update conda
conda create --name ml
conda activate ml

conda list
conda remove -n ml --all //remove environment

Cheatsheet
https://docs.conda.io/projects/conda/en/stable/_downloads/69c8c83c892884d9bc3700172a949a5b/conda-4.6.pdf

https://learn.microsoft.com/en-us/python/api/overview/azure/ai-ml-readme?view=azure-python
conda install python=3.7 //Use without the version, just conda install python to install 3.11

conda install -c microsoft azure-ai-ml //does not work, use conda pip

//No conda required here, environment is activated on the left of prompt pip install azure-ai-ml pip install azure-identity

Test with python script to try to authenticate with Azure

from azure.ai.ml import MLClient
from azure.identity import DefaultAzureCredential

ml_client = MLClient(
    DefaultAzureCredential(), subscription_id, resource_group, workspace
)

Where subscription_id, resource_group, workspace should be strings with double quotations copied from azure subscription and azure workspace.

https://learn.microsoft.com/hr-hr/azure/machine-learning/how-to-configure-environment?view=azureml-api-2

conda install notebook ipykernel //Enable environment-specific IPython kernels

ipython kernel install --user --name ml --display-name "Python (ml)" //Create a kernel for your Python virtual environment.

Launch the Jupyter Notebook server https://docs.anaconda.com/free/anaconda/jupyter-notebooks/remote-jupyter-notebook/
jupyter notebook --no-browser --port=8080