Table 6.
Confusion matrix in percentage for grading in two steps ({1&2} versus {3} versus {4} and then {1} versus {2}).
| % | SVM (Gini) | RF (Gini) | Grader B | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nb img | 279 | 279 | 176 | |||||||||
| Grader A | 1 | 2 | 3 | 4 | 1 | 2 | 3 | 4 | 1 | 2 | 3 | 4 |
|
| ||||||||||||
| 1 | 22.6 | 4.3 | 1.1 | 0.3 | 21.9 | 5.0 | 0.7 | 0.7 | 31.2 | 9.5 | 0.6 | 0.0 |
| 2 | 4.3 | 18.3 | 4.3 | 1.4 | 4.7 | 19.7 | 2.5 | 1.4 | 4.5 | 19.3 | 6.2 | 0.6 |
| 3 | 1.8 | 4.7 | 12.2 | 4.6 | 3.6 | 7.1 | 10.0 | 2.5 | 0.0 | 3.4 | 7.4 | 2.8 |
| 4 | 0.7 | 1.1 | 5.0 | 13.3 | 1.1 | 1.8 | 4.7 | 12.5 | 0.0 | 1.1 | 1.1 | 13.1 |
|
| ||||||||||||
| Accuracy | 66.3 | 64.2 | 71.5 | |||||||||
|
| ||||||||||||
| Weighted K (95% CI) | 66.2 (59.7–72.6) | 61.0 (53.8–68.1) | 73.6 (66.1–80.2) | |||||||||
| Substantial | Substantial | Substantial | ||||||||||