fhswf / mlpro-int-scikit-learn

MLPro: Integration scikit-learn

MLPro-Int-scikit-learn - Integration of scikit-learn into MLPro

Welcome to MLPro-Int-scikit-learn, an extension to MLPro to integrate the scikit-learn package. MLPro is a middleware framework for standardized machine learning in Python. It is developed by the South Westphalia University of Applied Sciences, Germany, and provides standards, templates, and processes for hybrid machine learning applications. Scikit-learn, in turn, provides numerous state-of-the-art algorithms for a vast amount of machine learning topics.

MLPro-Int-scikit-learn provides wrapper classes that enable the use of scikit-learn algorithms and data streams in your MLPro applications. The use of these wrappers is illustrated in various example programs.

Learn more

MLPro - Machine Learning Professional
MLPro-Int-scikit-learn - Integration of scikit-learn into MLPro
scikit-learn - Machine Learning in Python

mlpro-int-scikit-learn's Issues

OA-Streams-CA: Wrappers for top 5 online cluster analyzers

Description/Motivation
Scikit-learn is a very spread and mature ML package. Implementing wrappers for the same CA algorithms as we did for River enables statements of performance, accuracy, and reliability in a direct comparison...

This is motivated by the fact, that the current River implementations of STREAMKMeans, CluStream, DenStream, DBStream don't perform satisfyingly due to unknown reasons. A comparison with scikit-learn algos enables discovering the reasons...

A set of stable top 5 ca algorithms is also a good basis for further studies on ca-based anomaly detection.

Task list

• 1. Wrapper classes
  • KMeans
  • STREAMKMeans
  • CluStream
  • DenStream
  • DBStream
• 2. Related Howtos
  • Proposal: copy/paste/adjust the howtos from MLPro-Int-River
• 3. RTD
  • Proposal: copy/paste/adjust the related sections from MLPro-Int-River

Related issues
#...

Cross references
...

Branch
oa/cluster_analysis

Refact: Anomaly Detection - Wrappers, Howtos, RTD

To do

• Refactoring Howtos oa_ad_*
  Currently, data streams from OpenML are used. This dependency shall be removed. Instead, the new native stream Point Outliers shall be used.
• Wrappers
  Review, validation, refactoring
• RTD Updates
  • Integration Howtos (Detlef)
  • Completion API (class diagrams) (Syam)

See also
mlpro/test/bf/howto_bf_streams_005_native_stream_PointOutliersND.py

Branch
oa/anomaly_detection

Initial setup

Description/Motivation
After relocating the related wrapper and howto code the following steps are required.

Task list

• 1. Setup workflow ci.yml (Detlef)
• 2. Integration RTD (Detlef)
• 3. Integration PyPI (Detlef)
• 4. Integration Conda (Detlef)
• 5. Completion of RTD
  • Introduction
  • API
  • Howtos
• 6. Release v0.1.0
• 7. Integration MLPro extension hub (follow steps in auto-generated issue)