Actual class | Nb pigs | Percent correct | Predicted classes |
|---|
High RFI | Low RFI |
|---|
Random Forest procedure |
High RFI | 38 | 94.7% | 36 | 2 |
Low RFI | 36 | 97.2% | 1 | 35 |
Total | 74 | | |
Overall %Correct | 96.0% |
Gradient Tree Boosting procedure |
High RFI | 38 | 100% | 38 | 0 |
Low RFI | 36 | 100% | 0 | 36 |
Total | 74 | | |
Overall %Correct | 100% |
- Random forest (RF) and gradient treenet boosting (GTB) algorithms were applied on transcriptomic dataset from the whole blood sampled from 148 pigs of lines divergently selected for residual feed intake (RFI). Pigs were randomly split into training (n = 74) and validation test (n = 74) datasets to evaluate model performance in classifying pigs into low or high RFI groups. Expression levels of 50 molecular probes were considered in the validation set. The model made no error (100% of success) when built by GTB procedure
