Web Application and API for Association Rule Learning, Classification and Anomaly Detection
EasyMiner (EasyMiner/R) is an open source web-based visual interface and REST API for association rule learning. The R version of EasyMiner uses the fast apriori implementation in C from Christian Borgelt, as made available in the arules package in R. The system features implementation of the Classification Based on Associations (CBA) algorithm, which can be used for building classification models from association rules as well as for rule pruning, which addresses the common problem of too many discovered rules. EasyMiner/R also offers an experimental REST-based Prediction API. Unique to EasyMiner is interactive interface that allows to easily define a pattern for rules that you are looking for in your dataset. EasyMiner processes attribute-value data, rather than data in the transaction format.
Demo
News
- Editable Machine Learning Models? A rule-based framework for user studies of explainability published in Journal of Advances in Data Analysis and Classification
Preprint: https://nb.vse.cz/~klit01/papers/RuleEditor.pdf
Publisher website: https://link.springer.com/article/10.1007%2Fs11634-020-00419-2
- HAHSLER, Michael, JOHNSON, Ian, KLIEGR, Tomáš, KUCHAŘ, Jaroslav. Associative Classification in R: arc, arulesCBA, and rCBA. R Journal. 2019. 14 s. ISSN 2073-4859.
- The awarded paper is "Jiri Filip, and Tomas Kliegr. "PyIDS–Python Implementation of Interpretable Decision Sets Algorithm by Lakkaraju et al, 2016⋆." RuleML Challenge (2019)."
- This new algorithm is available on GitHub as pyIDS Python package.
- It is possible to use whole EasyMiner data mining and rule editor workflow also as new, simplified workflow suitable for crowdsourcing experiments. An experimenter can define an experiment based on a prepared rule set and distribute it among participants using a specific URL. The participants can use rule editor and model testing without creating own user accounts, all their activities are also logged for a later analysis by the experimenter.
- These new features are available in EasyMiner v2.6.
- Show screencast
- Find this release on GitHub or test it on our development server.
- Possibility to control the main part of the EasyMiner (Mining UI) via smaller and touch-sensitive displays
- Whether you have a touch screen on your laptop or prefer to use a tablet, the use of the system will be easier and more convenient.
- Support for writing analytical reports in WordPress
- New plugin for CMS WordPress will provide the functionality for saving of task reports in WordPress. These reports are made available through the web. It is also possible to simply select parts of task reports and reuse (insert) them into analytical reports created as posts or pages in WordPress.
- Similar functionality has already been tested (in a old version of CMS Joomla!) in one of the older projects prior to EasyMiner, but the new implementation is easier to use and supports the most popular CMS.
- How to use it:
- Copy rules to Knowledge base
- Click on and edit the rule list
- Re-evaluate the model with
- The new Rule Editor was first showcased at ECDA'18 in Paderborn
- Using EasyMiner API for Financial Data Analysis in the OpenBudgets.eu Project
- Outlier (Anomaly) Detection Modelling in PMML Papers are freely downloadable from CEUR-WS!