Table 2.
Sensitivity (the proportion of lost calves correctly identified), specificity (the proportion of surviving calves correctly identified) and accuracy (the proportion of correct predictions) of the population-level (PBM) and individual-based (IBM) methods for predicting calf survival across all data sets. For the PBM, the estimated threshold value (186.5 m/h) and the bounding values of its 95% bootstrap confidence interval are shown. For 2012 data, we assumed the predicted calving status was true and therefore included all females predicted to have calved
| Performance measure | Data set | PBM | IBM | ||
|---|---|---|---|---|---|
| 153.0 m/h | 186.5 m/h | 249.5 m/h | |||
| Sensitivity | 2011 (n = 4[3])1 | 1.0 | 0.75 | 0.25 | 1.0 |
| 2012 (n = 8) | 0.63 | 0.63 | 0.38 | 0.88 | |
| 2004 (n = 4) | 0.75 | 0.50 | 0.25 | 0.75 | |
| Specificity | 2011 (n = 8) | 0.63 | 0.75 | 1.0 | 0.75 |
| 2012 (n = 4) | 1.0 | 1.0 | 1.0 | 1.0 | |
| 2004 (n = 5) | 0.80 | 0.80 | 1.0 | 0.40 | |
| Accuracy | 2011 (n = 12) | 0.75 | 0.75 | 0.75 | 0.82 |
| 2012 (n = 12) | 0.75 | 0.75 | 0.58 | 0.92 | |
| 2004 (n = 9) | 0.78 | 0.67 | 0.67 | 0.56 | |
The total number of calves known to be lost in 2011 is three for IBM as we excluded the female misclassified as not calving.