Different stages of improvement of automatically arranged labels for self organizing map units.
The old, confusing way of labeling the winning neurons.
7. Mai 2007, 21:51: Labels are grouped by neurons now. This means that all organism a neuron responds to are written as a block with just one line to the neron on the map.
7. Mai 2007, 22:22: Labels do not overlap now, but only arranged based on the position of the neuron and then optimized along a pattern resembling a star.
8. Mai 2007, 03:56: Labels are now arranged on an ellypse arond the median around all responding neurons and then optimized on their circular angles.
8. Mai 2007, 15:06: Better overlap-detection for label-lines and labels, improved optimisazion by altering angle as well as radius. But there is still a problem with the map dimension! It should be 15x10 but is actualy 16x11.
9. Mai 2007, 14:54: Fixed map dimensions and plot size. Fine-tuned the label positioning, title positioning and label-lines.
Without label-boxes but with median-elypse displayed, which both can be turned on or off in the GUI. The median is 50 pixels shifted to the left because the labels fit better that way as the text goes from left to right.
10. Mai 2007, 02:12: Improved distribution of labels to be more use the map more efficiently by trying to find free space in a wave-like search path.
The wave-like search path can optionally be shown in the GUI to make clear why the labels at their respective postions.
