Skip to main content
. 2017 Feb 17;8:25. doi: 10.3389/fendo.2017.00025

Table 4.

The accuracy obtained as a result of training the ensemble of C5.0 boosted decision trees.

Trial no. Decision tree size Errors
0 11 6 (2.4%)
1 6 16 (6.4%)
2 4 49 (19.6%)
3 7 17 (6.8%)
4 8 11 (4.4%)
5 9 8 (3.2%)
6 9 16 (6.4%)
7 7 21 (8.4%)
8 8 28 (11.2%)
9 14 4 (1.6%)
Boosted ensemble 0 (0.0%)

Notice the error rate decreases as more decision trees are inducted into the ensemble, with a final (training) error of 0, i.e., resulting in a perfect classification of all training examples.