Model Performance Comparison.

BER, Kappa coefficients and Accuracy statistics calculated on the test set. Values indicated as being significantly better than the baseline (p≤0.05) are indicated in bold. Model numbers indicate the input features used; 1 indicates primary features; 2 indicates subset of features chosen by Boruta; 3 indicates all input features were used. K-NN1 is not included as the LOOcv indicated that 1 nearest neighbour gave the best performance making it the same as the baseline 1-NN model. (NB = Naive Bayes; RF = Random Forest; CT = Classification Tree; NN = Neural Network; SVM = Support Vector Machine; k-NN = k-nearest neighbour).