|
Figure 8
Classifier validation plots. (a) Confusion matrix for a subset of the complete testing data set classified as `high' resolution (≤4 Å). (b) Confusion matrix for a subset of the complete training data set classified as `low' resolution (>4 Å). (c) Receiver operating characteristic (ROC) curve for the resolution groups. (d) Classifier validation metrics for each resolution group, comparing overall ROC AUC, F1 score, recall, precision and accuracy. ROC AUC, receiver operating characteristic area under the curve. |