Fig. 4

a Percentage prediction accuracies and precisions for different classification models, based on PCR test results as reference. Models compared included k-nearest neighbours (KNN), logistic regression (LR), support vector machines (SVM), naïve Bayes (NB), XGBoost (XGB), random forest (RF), Multilayer perceptron (MLP). Logistic regression (LR) was the best performing model; b distribution of per class accuracies obtained by final LR classifiers and standard deviation from 70 bootstrapped models in predicting PCR test results from MIR spectral data. In both figures, accuracy refers to the how high the percentage prediction for each individual classification method is, while precision implies the statistical variation around those predictions