The inspection system gave accuracies of 94%and87%7% for wholesome and unwholesome carcasses, respectively.
All the machine learning algorithms gave accuracies as high as or higher than any of the semi-quantitative methods.
In many cases, using very few PC (usually less than 50) gave accuracies very close to the maximum obtained across the whole range of number of PC included.
In fact, this method, which was intentionally chosen as a 'naive' one, gave accuracies that were barely better than G BASE and predictions were still biased.
The re-substitution test on the training set gave accuracies of 100%, 100%, 0%, 100%, 66.66% and 100% for Classic, PlusC, Atypical1, Atypical2, Atypical3 and Atypical4 classes, respectively.
To summarize, BayesRS gave accuracies comparable to, but not always higher than or significantly different from, a simple pooling of the data.
In both 10 K SNP simulated data and 630 K real dairy cattle data, emBayesR gave accuracies of genomic prediction similar to BayesR, with greatly reduced computing time.
Speedups over traditional methods for a given accuracy are reported.
Thus, for given accuracy, model, and specimen size, minimum measurements duration can be determined.
