Abstract | ||
---|---|---|
•A new kind of diagrams is proposed, called phi-delta diagrams.•These diagrams are mainly aimed at supporting classifier building and assessment.•Accuracy, bias, and further relevant information are immediately evident therein.•Two kinds of diagrams have been devised: 1. not biased by the imbalance, 2. biased.•The former are an alternative to ROC curves, whereas the latter to coverage plots. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1016/j.ins.2018.06.052 | Information Sciences |
Keywords | Field | DocType |
Classifier performance measures,Confusion matrices,ROC curves,Coverage plots | Receiver operating characteristic,Intelligent decision support system,Design choice,Feature ranking,Visual tool,Software,Artificial intelligence,Phenomenon,Classifier (linguistics),Machine learning,Mathematics | Journal |
Volume | ISSN | Citations |
463 | 0020-0255 | 1 |
PageRank | References | Authors |
0.36 | 21 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Giuliano Armano | 1 | 325 | 42.89 |
Alessandro Giuliani | 2 | 170 | 25.21 |