Free Access
Table 4.
Definition of the terms and the metrics that are used in this work for the valauation of the classifiers.
| Term | Definition | Formula |
|---|---|---|
| True positive (TP) | An object is predicted from the classifier to belong to a spectral class, and it actually belongs to this class. | |
| True negative (TN) | An object is predicted from the classifier to not belong to a spectral class, and it actually does not belong to this class. | |
| False positive (FP) | An object is predicted from the classifier to belong to a spectral class, and it actually does not belong to this class. | |
| False negative (FN) | An object is predicted from the classifier to not belong to a spectral class, and it actually belongs to this class. | |
| Accuracy | Number of objects that are predicted correctly from the classifier, over the total number of the tested sample | (TP+TN) / (TP+TN+FP+FN) |
| Precision | Number of objects that are predicted correctly from the classifier to belong to a spectral class, over the total number of objects predicted by the classifier to belong to this class. | TP / (TP+FP) |
| Recall | Number of objects that correctly predicted from the classifier to belong to a spectral class, over the total number of objects that actually belong to this class. | TP / (TP+FN) |
| F1-score | The harmonic mean of the Recall and Precision. It is as an overall performance metric for the classifiers | 2TP / (2TP+FP+FN) |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.