Answers are grounded in this paper and cite their sources.
The aim of the present study was to illustrate the predictive performance of random forest (RF) used for dystocia detection in dairy cattle. A total of 1,342 and 1,699 calving records of Polish Holstein-Friesian Black-and-White heifers and cows were used. Five or ten predictor variables were included in the RF models for heifers and cows, respectively. The output variable was calving class. The proportion of correctly detected easy, moderate and difficult calving events in heifers on the independent test set was 39.64 %, 57.39 % and 83.64 %, respectively. The total accuracy was recorded as 60.12 %. The corresponding values for cows were 69.39 %, 67.61 %, 0 % and 66.04 %. The most significant predictors for heifers were sire’s rank and calving age, whereas those for cows additionally included: daily milk yield for the preceding lactation and the length of calving interval. The RF model developed in the present study was characterized by a high percentage of correctly diagnosed difficult calving events in heifers. However, it was completely unable to correctly detect dystocia in cows. The use of more influential predictor variables for cows in future research is especially important.
No takes yet. Share an insight, caveat, or question.
Zaborski et al. (2017) studied this question.
Synapse has enriched 6 related papers. Consider them for comparative context: