Skip to main content
. Author manuscript; available in PMC: 2015 Feb 24.
Published in final edited form as: Neuroinformatics. 2011 Sep;9(0):233–245. doi: 10.1007/s12021-011-9117-y

Fig. 8.

Fig. 8

Relationships between scoring methods. Data sets are identified with different symbols and abbreviations (HC: hippocampal CA3 interneuron; OP: olfactory projection fibers; NC: neocortical layer 6 axons; CF: cerebellar climbing fibers). All scores were normalized to a scale of 0 (incorrect) to 1 (perfect). A) Separation of scores by data set clearly shows the effect of data set on interaction times. B) The lower DIADEM metric scores (along the left edge) are responsible for much of the lack of correlation. These scores are largely due to terminal branches that track the gold standard well, but terminate in the wrong location. The effect is much stronger in the data sets containing trees with fewer branches (HC and NC).