Azure Machine Learning
An Azure machine learning service for building and deploying models.
Why is an Azure Machine Learning tutorial notebook portion not running successfully?
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(128, 8%, 22%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EA%3C/text%3E%3C/svg%3E)
Anonymous
I started following along with the explore-data.ipynb notebook found under Samples > SDK v2 > tutorials > get-started-notebooks. At code cell [5] I am running into an error.
Here is the block of code:
import pandas as pd
# get a handle of the data asset and print the URI
data_asset = ml_client.data.get(name="credit-card", version=v1)
print(f"Data asset URI: {data_asset.path}")
# read into pandas - note that you will see 2 headers in your data frame - that is ok, for now
df = pd.read_csv(data_asset.path)
df.head()
The data asset path prints fine on line 5. However, the second-to-last line pd.read_csv(data_asset.path) raises an error (below):
RuntimeError: StreamError(PermissionDenied(Some(AuthenticationError("AttributeError: 'MLClient' object has no attribute '_base_url'"))))
The kernel is set to Python 3.10 - SDK V2.
Having followed the tutorial steps, I'm wondering whether something has changed (perhaps with read_csv) since the tutorial was released, if something about the workspace configuration needs to be adjusted for the code to run successfully, or if it's some other issue. What can be done?
Azure Machine Learning
{count} votes
-
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator2023-08-30T07:21:51.2066667+00:00 @Anonymous It should work if the file is uploaded to the workspace blob store. See the screen shot of the same:
Just curious, have you run the steps mentioned in the first cell of my screenshot by giving the correct path to the local file to upload?
-
Anonymous
2023-08-30T13:25:25.64+00:00 Thank you for your response, @romungi-MSFT . Yes, on a previous run the data asset was created. You can see the created data asset in the screenshot:
Also, lines 4 and 5 run successfully for me (they retrieve and print the URI of the data asset). The code does not fail until I reach the
pd.read_csv(data_asset.path)step. Any further thoughts on what could be the issue? -
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator
2023-08-30T14:57:13.4566667+00:00 @Anonymous I am using the same kernel too. I have printed the _base__url value of the mlclient object and it seems to set it to 'https://management.azure.com'
I don't think this is set in this code snippet. This has to be part of the SDK, I feel that something is wrong with your SDK setup or some installation failed. Try to remove and reinstall the same. I am using SDK version 1.48.0 Thanks!!
-
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator
2023-09-01T07:55:56.5933333+00:00 @Anonymous Did the reinstall of the SDK help in your case?
-
Anonymous
2023-09-01T13:10:52.24+00:00 @romungi-MSFT I have been unsure how to proceed with uninstalling and reinstalling the SDK. I started searching but haven't found something yet that I was sure was what you meant. If you can provide an article or video or some detailed instructions yourself, I'd appreciate it!
-
romungi-MSFT 49,096 Reputation points Microsoft Employee Moderator
2023-09-03T15:25:39.9033333+00:00 @Anonymous Try a pip uninstall and install of the following packages in your notebook environment for the SDK.
%pip uninstall azure-ai-ml %pip uninstall azure-identity %pip install azure-ai-ml %pip install azure-identity
Sign in to comment
Sign in to answer