Table 3

Algorithm performance when comparing to expert maps: accuracy, precision, recall, f1-measure and True Skill Statistic (TSS) for each classifier. The first column under each metric shows scores when only EUV images were used, while the second column shows scores when Hα was included. “G” denotes the Gaussian model of Rigler et al. (2012); “M” is our Gaussian mixtures model; “F” is our random forest model. At the bottom of the table, the average row shows the overall performance, with each class equally weighted. The entries highlighted green indicate which classifier and image set combination performed best on that metric.