Fig. 8From: A comparison of machine learning and Bayesian modelling for molecular serotypingResults of combining two methods to divide the dataset into two groups of samples. Combining two methods to divide the dataset into two groups of samples, a high confidence group of samples (Group C) and others (Group NC), one method being the agreement between the Bayesian model and the GBM, the other method being a threshold on the percentage of a serotype’s cps genes significantly present. Graph plots the error rate in Group C on the left y-axis and the proportion of the dataset in Group C on the right y-axis, against the threshold used in the second method. For datasets D.36 and D.73Back to article page