Skip to main content
. 2016 Dec 5;36(1):255–268. doi: 10.1007/s00338-016-1519-8

Table 2.

Random forest classification training model results for (a) Percentage live coral cover, out-of-bag (OOB) estimate of error rate: 42.5%, (b) presence (1)/absence (0) of live coral cover within the microbathymetry, OOB estimate of error rate: 21.95% (c) presence (1)/absence (0) of live coral cover within the microbathymetry with high-definition samples, OOB estimate of error rate: 18.18%

(a) Percentage coral cover Class.error
=0 ]0;25] ]25;50] ]50;75] ]75;100]
=0 19 4 1 2 0 0.269
]0;25] 5 10 4 2 0 0.524
]25;50] 0 2 8 5 0 0.466
]50;75] 1 3 4 9 0 0.470
]75;100] 0 1 0 0 0 1.000
(b) 1 0 Class.error
1 49 5 0.070
0 14 12 0.560
(c) 1 0 Class.error
1 58 5 0.079
0 11 14 0.440

Number of trees = 1500, number of variables per level = 6