aws / aws/sagemaker-python-sdk
Feature Store Iceberg DataSource not implemented
- Dominant language
- Python
- Stars
- 2.3k
- Forks
- 1.3k
- Avg merge
- 1d 22h
- Merged PRs (30d)
- 35
Description
**Describe the feature you'd like**
IcebergDataSource is only partially implemented (see [_data_sources.py](https://github.com/aws/sagemaker-python-sdk/blob/32dd6314c4efb66983d17183e832502933cd23e3/src/sagemaker/feature_store/feature_processor/_data_source.py#L140)). This should be implemented, or the parameters and methodology for implementing it and dropping it in as a consumer of the library should be clearly documented.
**How would this feature be used? Please describe.**
The IcebergDataSource would just be used in any Feature Store step as any other data source, and would read my input data as defined in the object and passed to my Feature Store configuration. It would correctly construct the data frame from the specified Iceberg Table.
**Describe alternatives you've considered**
N/A
**Additional context**
I have a simple implementation of this in a private fork I can submit for reference. It probably isn't up to standard but it's a point to start from.
Contributor guide
Assessment
This issue has not been assessed yet.