ClustEval documentation¶
ClustEval is an automated framework for cluster analysis written in Java which comes together with a Ruby on Rails website. This includes most common subtasks of every cluster analysis, such as
- data preprocessing & data normalization
- data format conversion
- clustering method execution
- parameter training and
- visualizations
Getting Started¶
The Framework¶
Our framework clusteval is intended to perform automatized cluster analysis of arbitrary data sets and clustering methods. The goal is, that any clustering method known to the framework can be applied to any known data set (with certain exceptions, partly imposed by the employed clustering methods and partly by ClustEval).
In general clusteval is divided into a backend and a frontend. Figure 1 shows the general structure of the framework. The backend is reponsible for doing all the calculations including clusterings and the frontend has only visualization purposes.
- ClustEval Server
- Data Sets
- Gold Standards
- Clustering Methods
- Configuration Files - Handling ClustEval
- Formats - Input & Output
- Clustering Quality Measures - Is It a Good Clustering?
- Distance Measures - Converting Absolute Coordinates to Pairwise Similarities
- Parameter Optimization Methods - Finding Good Clustering Parameters
- Data Set Types - Classifying Data
- Statistics - Analyzing Data & Clusterings
- Data Set Generators - Validation On Artificial Data
- Data Preprocessors - Cleaning Data and Removing Noise
- Data Randomizer - Distorting Existing Data Sets
- Repository
- CLI (Command line interface)
- ClustEval Client
- Website
- Database
- Web Administration Interface
Extending ClustEval¶
ClustEval can be extended in many ways. Please head over to Extending ClustEval to get more information.