3.5.1.4. 3rd Party Support

MLPro’s stream sub-package is a comprehensive package that supports third-party platforms in addition to its native stream object implementation. Third-party platforms provide access to various datasets, enabling to design, test, and validate stream processing workflows and online machine learning algorithms. These thrird party platforms are called StreamProvider in context of Stream sub-package of MLPro. Currently the streams sub-package supports following 3rd party packages:

3.5.1.4.1. OpenML

OpenML is a collaborative platform that provides access to a vast range of datasets and algorithms for machine learning. OpenML’s datasets are typically larger in size and contain more complex data, enabling more accurate results. MLPro supports OpenML’s datasets as stream objects.

These datasets are provided as streams containing metadata information about the datasets, including their names, descriptions, and the attributes they contain.

Learn More: OpenML

Note

Unfortunately, there have been some issues with OpenML’s API integration in MLPro, causing it to not work as intended in some cases. Users may encounter errors or limitations when using OpenML data sources through MLPro’s stream provider functionality. It is recommended to keep an eye out for updates.

Note

MLPro’s stream provider functionality not only allows users to access third-party platforms but also provides additional options for loading stream objects from these platforms. For instance, when accessing datasets from OpenML, MLPro users can choose between default or custom input and target features.

You can import OpenML Stream Provider wrapper as following:

# Importing OpenML Stream Provider
from mlpro.wrappers.openml import WrStreamProviderOpenML

Cross References

• Howto BF-STREAMS-051: Accessing Data from OpenML

• Wrapper OpenML

3.5.1.4.2. Scikit-learn

Scikit-learn is a popular and well-established machine learning package that provides various algorithms for classification, regression, clustering, and dimensionality reduction. ScikitLearn also provides a wide range of datasets which includes popular examples such as the Iris dataset for classification, and the California Housing dataset for regression that are frequently used for training and testing machine learning models. MLPro currently supports scikit-learn’s datasets as stream objects, which can be used in stream processing and online machine learning.

Learn more: Scikit-learn

You can import scikit-learn Stream Provider wrapper as following:

# Importing scikit-learn Stream Provider
from mlpro.wrappers.sklearn import WrStreamProviderSKLearn

Cross References

• Howto BF-STREAMS-052: Accessing Data from Scikit-Learn

• Wrapper Scikit-learn

3.5.1.4.3. River

River is a machine learning library designed for streaming data. It is used for real-time analytics and online machine learning where data is constantly evolving. River provides various online learning algorithms that are optimized for processing streaming data. MLPro supports River’s datasets as data sources, which can be used to build machine learning models that are optimized for streaming data.

River datasets include specialized datasets for online machine learning which is commonly used for evaluating online learning algorithms. MLPro standardizes the access to the River streams, making it easier to incorporate them into stream processing workflows. While MLPro currently supports River as a data resource, future updates will include standardized support for algorithms in River.

Learn more: River ML

You can import River Stream Provider wrapper as following:

# Importing River Stream Provider
from mlpro.wrappers.river import WrStreamProviderRiver

Learn more in the Howto section of MLPro, in appendix 01 section.

Cross References

• Howto BF-STREAMS-053: Accessing Data from River

• Wrapper River