Azure Machine Learning distinguishes between uri_folder, uri_file, and mltable data assets. Because the registered asset in this scenario is a folder data asset containing CSV data , the appropriate MLTable API is mltable.from_delimited_files(). Microsoft documents that a registered uri_folder containing CSV files can be retrieved through MLClient, its storage path supplied as a folder path, and then passed to mltable.from_delimited_files(). The resulting MLTable object can subsequently be materialized into a Pandas DataFrame by calling to_pandas_dataframe().

A representative workflow is conceptually:

tbl = mltable.from_delimited_files(paths=[{ " folder " : data_asset.path}])

followed by:

df = tbl.to_pandas_dataframe()

mltable.load() is primarily appropriate when loading an existing MLTable definition , such as a registered table data asset containing an MLTable file. from_parquet_files() is intended for Parquet-formatted data, while from_delta_lake() targets Delta Lake data sources. Neither matches the CSV-based folder described here. Microsoft also identifies uri_folder as the appropriate asset type for folders containing CSV or Parquet files used with Pandas or Spark.

Study Guide Reference: Design and implement an MLOps infrastructure — Azure Machine Learning data assets, MLTable APIs, interactive data access, and Pandas-based data wrangling.

Continue with Azure ML data access