Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations]
Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews i...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
F1000 Research Ltd
2017-02-01
|
| Series: | F1000Research |
| Subjects: | |
| Online Access: | https://f1000research.com/articles/5-2762/v3 |
| _version_ | 1818326568458518528 |
|---|---|
| author | Ignacio Enrique Sanchez |
| author_facet | Ignacio Enrique Sanchez |
| author_sort | Ignacio Enrique Sanchez |
| collection | DOAJ |
| description | Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs. |
| first_indexed | 2024-12-13T12:02:27Z |
| format | Article |
| id | doaj.art-61a6571dd5ba49c3b62f73cc84e5a34d |
| institution | Directory Open Access Journal |
| issn | 2046-1402 |
| language | English |
| last_indexed | 2024-12-13T12:02:27Z |
| publishDate | 2017-02-01 |
| publisher | F1000 Research Ltd |
| record_format | Article |
| series | F1000Research |
| spelling | doaj.art-61a6571dd5ba49c3b62f73cc84e5a34d2022-12-21T23:47:04ZengF1000 Research LtdF1000Research2046-14022017-02-01510.12688/f1000research.10114.311399Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations]Ignacio Enrique Sanchez0Protein Physiology Laboratory, University of Buenos Aires, Buenos Aires, ArgentinaMany bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs.https://f1000research.com/articles/5-2762/v3Bioinformatics |
| spellingShingle | Ignacio Enrique Sanchez Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] F1000Research Bioinformatics |
| title | Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] |
| title_full | Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] |
| title_fullStr | Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] |
| title_full_unstemmed | Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] |
| title_short | Optimal threshold estimation for binary classifiers using game theory [version 3; referees: 2 approved, 1 approved with reservations] |
| title_sort | optimal threshold estimation for binary classifiers using game theory version 3 referees 2 approved 1 approved with reservations |
| topic | Bioinformatics |
| url | https://f1000research.com/articles/5-2762/v3 |
| work_keys_str_mv | AT ignacioenriquesanchez optimalthresholdestimationforbinaryclassifiersusinggametheoryversion3referees2approved1approvedwithreservations |