# Azure Machine Learning Feature Store Python SDK
The `azureml-featurestore` package is the core SDK interface for Azure ML Feature Store. This SDK works along the
`azure-ai-ml` SDK to provide the managed feature store experience.
## Main features in the `azureml-featurestore` package
- Develop feature set specification in Spark with the ability for feature transformation.
- List and get feature sets defined in Azure ML Feature Store.
- Generate and resolve feature retrieval specification.
- Run offline feature retrieval with point-in-time join.
## Getting started
You can install the package via ` pip install azureml-featurestore `
To learn more about Azure ML managed feature store visit https://aka.ms/featurestore-get-started
# Change Log
## 1.1.0 (2024.03.13)
**New Features:**
- [Public Preview] Support for DSL (Domain Specific Language) for feature definition. The DSL is a simplified way to define feature set transformations using a declarative syntax.
- DSL feature set supports custom source
- DSL feature set supports temporal join lookback, source delay and source lookback
- DSL feature set supports load from materialized store
- `get_offline_features supports` feature sets with different transformations(dsl, udf or none).
## 1.0.1 (2023.12.28)
- Update dependencies
## 1.0.0 (2023.11.14)
- [GA] Custom feature source: Custom feature source supports customized source process code script with a user defined dictionary as input.
- [GA] International regions and sovereign cloud support.
- [GA] Offline backfill materialization now replaces all data within a feature window instead of doing upsert based on timestamp.
- [GA] Added bootstrap option for materialization, which enables materializing data from offline store into online store.
- Re-enabling materialization in a feature set now invalidates all previously materialized data.
- Feature set spec dump now accepts a file path or a folder path as dump target, and an overwrite option to control whether to override the target.
- Various bug fixes
## 0.1.0b6 (2023.11.1)
- Various bug fixes
## 0.1.0b5 (2023.10.4)
- Various bug fixes
## 0.1.0b4 (2023.08.28)
**New Features:**
- [Public preview] Added custom feature source. Custom feature source supports customized source process code script with a user defined dictionary as input.
- [Public preview] Added csv feature source, deltatable feature source, mltable feature source, parquet feature source as new feature source experience. Previous feature source usage compatibility will be deprecated in 6 months.
- Bug fixes
**Breaking changes:**
- Moved `init_online_lookup`, `shutdown_online_lookup` and `get_online_features` out of FeatureStoreClient, and into the module as standalone functions.
- `get_online_features` contract changed from accepting (for the `observation_data` argument) and returning `pandas.DataFrame` to accepting (as the `observation_data` argument) and returning `pyarrow.Table`.
**Other changes:**
- Moved online feature store support into public preview.
## 0.1.0b3 (2023.07.10)
- Various bug fixes
## 0.1.0b2 (2023.06.13)
**New Features:**
- [Private preview] Added online store support. Online store supports materialization and online feature values retrieval from Redis cache for batch scoring.
- Various bug fixes
## 0.1.0b1 (2023.05.15)
**New Features:**
Initial release.
Raw data
{
"_id": null,
"home_page": "https://aka.ms/featurestore-get-started",
"name": "azureml-featurestore",
"maintainer": "",
"docs_url": null,
"requires_python": "<4.0,>=3.8",
"maintainer_email": "",
"keywords": "AzureMachineLearning,azure,feature store",
"author": "Microsoft Corporation",
"author_email": "azuresdkengsysadmins@microsoft.com",
"download_url": "",
"platform": null,
"description": "# Azure Machine Learning Feature Store Python SDK\n\nThe `azureml-featurestore` package is the core SDK interface for Azure ML Feature Store. This SDK works along the\n`azure-ai-ml` SDK to provide the managed feature store experience.\n\n## Main features in the `azureml-featurestore` package\n- Develop feature set specification in Spark with the ability for feature transformation.\n- List and get feature sets defined in Azure ML Feature Store.\n- Generate and resolve feature retrieval specification.\n- Run offline feature retrieval with point-in-time join.\n\n## Getting started\n\nYou can install the package via ` pip install azureml-featurestore `\n\nTo learn more about Azure ML managed feature store visit https://aka.ms/featurestore-get-started\n\n\n# Change Log\n\n## 1.1.0 (2024.03.13)\n\n**New Features:**\n\n- [Public Preview] Support for DSL (Domain Specific Language) for feature definition. The DSL is a simplified way to define feature set transformations using a declarative syntax.\n - DSL feature set supports custom source\n - DSL feature set supports temporal join lookback, source delay and source lookback\n - DSL feature set supports load from materialized store\n - `get_offline_features supports` feature sets with different transformations(dsl, udf or none).\n\n## 1.0.1 (2023.12.28)\n\n- Update dependencies\n\n## 1.0.0 (2023.11.14)\n\n- [GA] Custom feature source: Custom feature source supports customized source process code script with a user defined dictionary as input.\n- [GA] International regions and sovereign cloud support.\n- [GA] Offline backfill materialization now replaces all data within a feature window instead of doing upsert based on timestamp.\n- [GA] Added bootstrap option for materialization, which enables materializing data from offline store into online store.\n- Re-enabling materialization in a feature set now invalidates all previously materialized data.\n- Feature set spec dump now accepts a file path or a folder path as dump target, and an overwrite option to control whether to override the target.\n- Various bug fixes\n\n## 0.1.0b6 (2023.11.1)\n\n- Various bug fixes\n\n## 0.1.0b5 (2023.10.4)\n\n- Various bug fixes\n\n## 0.1.0b4 (2023.08.28)\n\n**New Features:**\n\n- [Public preview] Added custom feature source. Custom feature source supports customized source process code script with a user defined dictionary as input.\n- [Public preview] Added csv feature source, deltatable feature source, mltable feature source, parquet feature source as new feature source experience. Previous feature source usage compatibility will be deprecated in 6 months.\n\n- Bug fixes\n\n**Breaking changes:**\n- Moved `init_online_lookup`, `shutdown_online_lookup` and `get_online_features` out of FeatureStoreClient, and into the module as standalone functions.\n- `get_online_features` contract changed from accepting (for the `observation_data` argument) and returning `pandas.DataFrame` to accepting (as the `observation_data` argument) and returning `pyarrow.Table`.\n\n**Other changes:**\n- Moved online feature store support into public preview.\n\n## 0.1.0b3 (2023.07.10)\n\n- Various bug fixes\n\n## 0.1.0b2 (2023.06.13)\n\n**New Features:**\n\n- [Private preview] Added online store support. Online store supports materialization and online feature values retrieval from Redis cache for batch scoring.\n\n- Various bug fixes\n\n## 0.1.0b1 (2023.05.15)\n\n**New Features:**\n\nInitial release.\n",
"bugtrack_url": null,
"license": "MIT License",
"summary": "Azure Machine Learning Feature Store SDK",
"version": "1.1.0",
"project_urls": {
"Homepage": "https://aka.ms/featurestore-get-started"
},
"split_keywords": [
"azuremachinelearning",
"azure",
"feature store"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "af7679c6c3351284708ef7c7aed92546e068125875d4a646a913ba210d053a90",
"md5": "57aecf1086c7004ea1bcfe07162afbfc",
"sha256": "d1ccb0df469925db2ef89e4556912c71d91ce831943e1674085395664d4a32ab"
},
"downloads": -1,
"filename": "azureml_featurestore-1.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "57aecf1086c7004ea1bcfe07162afbfc",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<4.0,>=3.8",
"size": 150891,
"upload_time": "2024-03-13T09:50:18",
"upload_time_iso_8601": "2024-03-13T09:50:18.618244Z",
"url": "https://files.pythonhosted.org/packages/af/76/79c6c3351284708ef7c7aed92546e068125875d4a646a913ba210d053a90/azureml_featurestore-1.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-03-13 09:50:18",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "azureml-featurestore"
}
JSON