Figure 2
Confusion matrices obtained after predicting the data points in the test set with each of the three classifiers being assessed: support-vector machines, random forest and k-nearest neighbours. Residues present in a register error are labelled as class 1 and residues modelled correctly are labelled as class 0. The numbers inside each cell correspond to the number of data points in each category, which is proportional to the colour intensity in each cell. |