Table 1.
Misclassification rates for binary classification trees differentiating between cattle grazing and non-grazing activities based upon synchronized GPS collar and direct observation data collected for yearling steers grazing shortgrass steppe in eastern Colorado during 2008–2011. The final selected model is shown in bold, and presented in Figure 2.
Number of Model Splits | r2 | Misclassification Rate | |||||
---|---|---|---|---|---|---|---|
| |||||||
Training Data | Validation Data | ||||||
|
|
||||||
Overall | False | Overall | False | ||||
|
|
||||||
Negatives | Positives | Negatives | Positives | ||||
4 | 0.505 | 14.3 | 13.8 | 15.0 | 14.1 | 15.4 | 12.2 |
Splitting within the (Headdown ≤ 94.7 %) category | |||||||
8 | 0.549 | 13.6 | 12.1 | 15.9 | 14.7 | 14.7 | 14.7 |
9 | 0.551 | 13.6 | 12.4 | 15.3 | 14.1 | 15.4 | 12.2 |
Splitting within the (Headdown ≥ 94.7 %) category | |||||||
6 | 0.512 | 14.0 | 13.0 | 15.5 | 13.4 | 13.6 | 13.0 |
9 | 0.529 | 13.8 | 13.2 | 14.6 | 12.9 | 13.5 | 12.2 |
11 | 0.530 | 13.7 | 13.4 | 14.1 | 13.2 | 14.0 | 12.2 |