Pavel

Pavel (Paretoset Analysis Visualization and Evaluation) is a tool for interactively displaying and evaluating large sets of highdimensional data. Its main intended use is the analysis of result sets from multi-objective evolutionary algorithms applied to machining problems but it can basically work with any kind of multidimensional data.

While there exist solutions that fulfill some of the listed purposes, none of them is combining them in a way that is useful for the evaluation of MOEA result sets. A unique feature of Pavel is further the ability to handle extremely large data sets well enough to enable interactive evaluation (At least up to 100.000 points in 40 dimensions). Pavel is easily extendable through plugin interfaces and can be applied to many different use cases.

The project was developed as a student research project at Dortmund University since in October 2006. When the project was nearing its end a year later, the software was usable but still left room for improvement. To make the effort spent worthwhile, it was decided to release the software as opensource in the hope that others may find it useful and to enable development to continue.

Currently (early 2008), development is driven by several university projects by members of the original development team.

Features

• Visualize large data sets in several ways (Scatterplots, Parallelplots, Scatter Matrices, Tables)
• Cluster data using different clustering algorithms (K-Means, HACM)
• Compare individual points in the data set
• Analyze data with algorithms related to the actual use case
• Extendable through plugin interfaces (applied to many different use cases)

Screenshot