Visualization in ELKI
Usage
To use the ELKI visualizer, set -resulthandler to de.lmu.ifi.dbs.elki.result.AutomaticVisualization. This class will inspect the algorithm output, invoke appropriate visualization modules and arrange them on the screen. The menu allows export to SVG, PNG, JPG, PDF, PostScript and other formats.
Technology
ELKI uses SVG for visualization (which has benefits for print export), and Apache Batik for screen rendering. While this is not the fastest solution, the export benefits are good and we don’t re-write pixel-rendering code ourselves.
See Batik for documentation pointers.
Visualization modules
Visualizers included in ELKI 0.6.0:
- Histogram visualizations:
- histogram.ColoredHistogramVisualizer (data histograms)
- visunproj.HistogramVisualization (histogram results)
- Scatterplots:
- Basic:
- Object Selection:
- Clustering:
- Outlier detection:
- Geo data:
- Specialized:
- scatterplot.density.DensityEstimationOverlay
- scatterplot.selection.DistanceFunctionVisualization (visualize kNN of selection)
- scatterplot.cluster.VoronoiVisualization (for K-Means)
- scatterplot.cluster.ClusterOrderVisualization (for OPTICS)
- scatterplot.ReferencePointsVisualization (Algorithm reference points used)
- scatterplot.index.TreeMBRVisualization (visualize R-Trees)
- scatterplot.index.TreeSphereVisualization (visualize M-Trees)
- Parallel coordinates:
- parallel.LineVisualization
- parallel.cluster.ClusterParallelMeanVisualization
- parallel.cluster.ClusterOutlineVisualization
- parallel.index.RTreeParallelVisualization
- parallel.ParallelAxisVisualization (axis rendering)
- parallel.selection.SelectionLineVisualization
- parallel.selection.SelectionAxisRangeVisualization
- OPTICS:
- optics.OPTICSPlotVisualizer
- optics.OPTICSClusterVisualization
- optics.OPTICSPlotCutVisualization (for epsilon cuts)
- optics.OPTICSPlotSelectionVisualization
- optics.OPTICSSteepAreaVisualization (for OPTICS Xi method)
- Pair segments visualization:
- Generic:
- Textual: