KEEL
sci2s.ugr.es
- Free plan
- Windows
- macOS
- Linux
What KEEL is
KEEL is a free Java software suite for data mining, intended for research and education. It covers regression, classification, clustering, and pattern mining, and its listed catalog includes 518 algorithms. Preprocessing methods include transformation, discretization, instance and feature selection, and noise filtering. Users can use the graphical interface to assemble experiments from multiple datasets and algorithms; KEEL then generates files for running those experiments offline on a machine with a Java Virtual Machine. The suite includes a statistical library with parametric and nonparametric tests for comparing results, as well as modules for imbalanced learning, semi-supervised learning, nonparametric statistical analysis, and multiple instance learning. It can convert datasets in CSV, ARFF, or plain-text formats into KEEL format. The download includes a basic dataset package and links to a repository with more than 600 datasets. KEEL is open source under GPLv3, and downloads include source code and a binary for release 3.0. The project says Windows, Linux, and Mac OS X are supported with Java 7 or later; the reference manual specifies a JVM version of 1.7 or later.
Who it suits
KEEL is suited to researchers and educators who want a local Java environment for data mining experiments and statistical comparisons. It may also fit users who need preprocessing methods and multiple learning modules in one suite.
Good to know, and what to watch
- Free software under GPLv3
- Covers regression, classification, clustering, and pattern mining
- Includes 518 listed algorithms
- Supports CSV, ARFF, and plain-text data conversion
- Runs on Windows, Linux, and Mac OS X
- Requires a Java Virtual Machine
- Experiments run offline on a local machine
- The available binary is release 3.0
Plans
Every plan and priceKEEL
Free plan
The verdict
KEEL offers a wide range of data-mining methods and experiment tools for local research or teaching workflows. Users should be comfortable installing a Java runtime and running experiments offline.
What the maker publishes
- Algorithms
- The site lists 518 algorithms, including data preprocessing and classification algorithms. source
- Audience
- The site describes KEEL as designed for research and education. source
- Data import
- The manual says KEEL can convert datasets from CSV, ARFF, or plain text formats into KEEL format. source
- Datasets
- The download page says the software includes a basic dataset package and links to a repository with more than 600 datasets. source
- Development group
- The project members page identifies the University of Granada group as the coordinating node for KEEL development. source
- Download
- The download page offers the source code and a binary version for release 3.0 dated 2018-04-09. source
- Experiment design
- Its graphical interface lets users design experiments with multiple datasets and algorithms and generates files to run experiments offline on a machine with a Java Virtual Machine. source
- Integration
- The download page links to documentation for integrating new algorithms into KEEL. source
- Modules
- The suite includes modules for imbalanced learning, nonparametric statistical analysis, semi-supervised learning, and multiple instance learning. source
- Preprocessing
- KEEL includes preprocessing methods for transformation, discretization, instance and feature selection, and noise filtering. source
- Purpose
- KEEL is a Java software suite for data mining tasks including regression, classification, clustering, and pattern mining. source
- Runtime
- KEEL requires a Java Virtual Machine; the reference manual says JVM version 1.7 or later should run the graphical interface and included algorithms. source
- Security and license
- The download page identifies KEEL as open-source software licensed under GPLv3. source
- Statistical analysis
- KEEL includes a statistical library and parametric and nonparametric tests for comparing algorithm results. source
- Supported systems
- The KEEL 3.0 paper states it can be used on Windows, Linux, and Mac OS X with Java 7 or later. source
Checked against the maker's pages on 5 Oct 2026.