Fig. 4: Evaluation metrics and performance comparison of the classification models. | npj Computational Materials

Fig. 4: Evaluation metrics and performance comparison of the classification models.

From: Interpretable X-ray diffraction spectra analysis using confidence evaluated deep learning enhanced by template element replacement

Fig. 4

a, b Confusion matrix, receiver operating characteristic (ROC), and precision-recall (PR) curve of the “Yes or No” model. c Structure classification accuracy across machine learning (ML) models verified on real structure spectral data (RSS), with color intensity representing accuracy levels. The vertical axis represents the training set, and the horizontal axis represents the model used.

Back to article page