scikit-multiflow is a machine learning package for streaming data in Python.

creme and scikit-multiflow are merging into a new project called River.

We feel that both projects share the same vision. We believe that pooling our resources instead of duplicating work will benefit both sides. We are also confident that this will benefit both communities. There will be more people working on the new project, which will allow us to distribute work more efficiently. We will thus be able to work on more features and improve the overall quality of the project.

Both projects will stop active development. The code for both projects will remain publicly available, although development will only focus on minor maintenance during a transition period. The architecture of the new package is very similar to that of creme. It will focus on single-instance incremental models.

We encourage users to use River instead of creme. We understand that this transition will require an extra effort in the short term from current users. However, we believe that the result will be better for everyone in the long run.

You will still be able to install and use creme as well as scikit-multiflow. Both projects will remain on PyPI, conda-forge and GitHub.

• Webpage
• Documentation
• Community

Stream learning models are created incrementally and are updated continuously. They are suitable for big data applications where real-time response is vital.

Changes in data distribution harm learning. Adaptive methods are specifically designed to be robust to concept drift changes in dynamic environments.

Streaming techniques efficiently handle resources such as memory and processing time given the unbounded nature of data streams.

scikit-multiflow is designed for users with any experience level. Experiments are easy to design, setup, and run. Existing methods are easy to modify and extend.

In its current state, scikit-multiflow contains data generators, multi-output/multi-target stream learning methods, change detection methods, evaluation methods, and more.

Distributed under the BSD 3-Clause, scikit-multiflow is developed and maintained by an active, diverse and growing community.

The following tasks are supported in scikit-multiflow:

When working with labeled data. Depending on the target type can be either classification (discrete values) or regression (continuous values)

Single-output methods predict a single target-label (binary or multi-class) for classification or a single target-value for regression. Multi-output methods simultaneously predict multiple variables given an input.

Changes in data distribution can harm learning. Drift detection methods are designed to rise an alarm in the presence of drift and are used alongside learning methods to improve their robustness against this phenomenon in evolving data streams.

When working with unlabeled data. For example, anomaly detection where the goal is the identification of rare events or samples which differ significantly from the majority of the data.

In order to display plots from scikit-multiflow within a Jupyter Notebook we need to define the proper mathplotlib backend to use. This is done by including the following magic command at the beginning of the Notebook:

%matplotlib notebook

JupyterLab is the next-generation user interface for Jupyter, currently in beta, it can display interactive plots with some caveats. If you use JupyterLab then the current solution is to use the jupyter-matplotlib extension:

%matplotlib widget

If scikit-multiflow has been useful for your research and you would like to cite it in a academic publication, please use the following Bibtex entry:

@article{skmultiflow,
  author  = {Jacob Montiel and Jesse Read and Albert Bifet and Talel Abdessalem},
  title   = {Scikit-Multiflow: A Multi-output Streaming Framework },
  journal = {Journal of Machine Learning Research},
  year    = {2018},
  volume  = {19},
  number  = {72},
  pages   = {1-5},
  url     = {http://jmlr.org/papers/v19/18-251.html}
}