Credit Card Fraud Detection on BankSim
77.2Partial PR AUCRandom Forest
Evaluation Results
| Method | Links | |
|---|---|---|
| Random ForestN_l=259462025.08 | 77.2 | |
| Fully supervised neural network (FNN)N_l=259462025.08 | 77.2 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesN_l=259462025.08 | 76.7 | |
| Random ForestN_l=129732025.08 | 75.4 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesN_l=129732025.08 | 75.2 | |
| Fully supervised neural network (FNN)N_l=129732025.08 | 74.2 | |
| FNK=0.8, Nl=259462025.08 | 72.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.8, Nl=259462025.08 | 72.3 | |
| RFK=0.8, Nl=259462025.08 | 71.7 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.8, Nl=129732025.08 | 71.4 | |
| Random ForestN_l=51902025.08 | 69.6 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesN_l=51902025.08 | 69.3 | |
| Support Vector MachineN_l=259462025.08 | 68.8 | |
| Random ForestN_l=38932025.08 | 68.3 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.8, Nl=51902025.08 | 67.2 | |
| SVMK=0.8, Nl=259462025.08 | 66.9 | |
| Fully supervised neural network (FNN)N_l=51902025.08 | 66.3 | |
| Support Vector MachineN_l=129732025.08 | 66.3 | |
| Random ForestN_l=25952025.08 | 66.1 | |
| FNlabeled training size (Nl)=259462025.08 | 66 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesN_l=38932025.08 | 65.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.7, Nl=259462025.08 | 65.7 | |
| Semi-Supervised Bayesian GANs with Log-Signatureslabeled training size (Nl)=259462025.08 | 65.7 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.7, Nl=129732025.08 | 65.1 | |
| Semi-Supervised Bayesian GANs with Log-Signatureslabeled training size (Nl)=129732025.08 | 65.1 | |
| FNlabeled training size (Nl)=129732025.08 | 64.9 | |
| RFlabeled training size (Nl)=259462025.08 | 64.9 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.8, Nl=38932025.08 | 64.2 | |
| RFlabeled training size (Nl)=129732025.08 | 63.7 | |
| Fully supervised neural network (FNN)N_l=38932025.08 | 62.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.7, Nl=51902025.08 | 62.4 | |
| Semi-Supervised Bayesian GANs with Log-Signatureslabeled training size (Nl)=51902025.08 | 62.4 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesN_l=25952025.08 | 62 | |
| SVMlabeled training size (Nl)=259462025.08 | 61.8 | |
| FNlabeled training size (Nl)=51902025.08 | 61.6 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.8, Nl=25952025.08 | 61 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.7, Nl=38932025.08 | 60.6 | |
| Semi-Supervised Bayesian GANs with Log-Signatureslabeled training size (Nl)=38932025.08 | 60.6 | |
| SVMlabeled training size (Nl)=129732025.08 | 60.3 | |
| RFlabeled training size (Nl)=51902025.08 | 60.2 | |
| RFlabeled training size (Nl)=38932025.08 | 59 | |
| FNlabeled training size (Nl)=38932025.08 | 59 | |
| Fully supervised neural network (FNN)N_l=25952025.08 | 58.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.7, Nl=25952025.08 | 58.2 | |
| Semi-Supervised Bayesian GANs with Log-Signatureslabeled training size (Nl)=25952025.08 | 58.2 | |
| Support Vector MachineN_l=51902025.08 | 58 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.6, Nl=259462025.08 | 57.4 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.6, Nl=129732025.08 | 56.9 | |
| RFlabeled training size (Nl)=25952025.08 | 56.9 | |
| FNlabeled training size (Nl)=25952025.08 | 56 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.6, Nl=51902025.08 | 55.3 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.6, Nl=38932025.08 | 54.4 | |
| SVMlabeled training size (Nl)=51902025.08 | 53.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.6, Nl=25952025.08 | 52.7 | |
| KNNlabeled training size (Nl)=259462025.08 | 51.3 | |
| Naive BayesN_l=51902025.08 | 50.5 | |
| Naive BayesN_l=129732025.08 | 50.5 | |
| Naive BayesN_l=259462025.08 | 50.4 | |
| Naive BayesN_l=38932025.08 | 50.3 | |
| Support Vector MachineN_l=38932025.08 | 50 | |
| Naive BayesN_l=25952025.08 | 49.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.5, Nl=259462025.08 | 48.2 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.5, Nl=129732025.08 | 47.9 | |
| KNNlabeled training size (Nl)=129732025.08 | 47.2 | |
| SVMlabeled training size (Nl)=38932025.08 | 47.1 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.5, Nl=51902025.08 | 47 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.5, Nl=38932025.08 | 46.6 | |
| NBlabeled training size (Nl)=38932025.08 | 45.9 | |
| NBlabeled training size (Nl)=51902025.08 | 45.9 | |
| NBlabeled training size (Nl)=129732025.08 | 45.9 | |
| NBlabeled training size (Nl)=259462025.08 | 45.9 | |
| NBlabeled training size (Nl)=25952025.08 | 45.8 | |
| Semi-Supervised Bayesian GANs with Log-SignaturesK=0.5, Nl=25952025.08 | 45.5 | |
| K-Nearest NeighborsN_l=259462025.08 | 44.3 | |
| FNK=0.5, Nl=25952025.08 | 44.1 | |
| RFK=0.5, Nl=25952025.08 | 43.4 | |
| Support Vector MachineN_l=25952025.08 | 41.7 | |
| KNNlabeled training size (Nl)=51902025.08 | 41.3 | |
| Semi-Supervised Logistic RegressionN_l=259462025.08 | 40.7 | |
| KNNlabeled training size (Nl)=38932025.08 | 39.9 | |
| SVMlabeled training size (Nl)=25952025.08 | 39.5 | |
| Logistic RegressionN_l=259462025.08 | 38.4 | |
| L-SSLlabeled training size (Nl)=259462025.08 | 38.3 | |
| NBK=0.5, Nl=25952025.08 | 37.7 | |
| KNNlabeled training size (Nl)=25952025.08 | 37.5 | |
| Llabeled training size (Nl)=259462025.08 | 36.3 | |
| K-Nearest NeighborsN_l=129732025.08 | 35.9 | |
| Semi-Supervised Logistic RegressionN_l=129732025.08 | 33.8 | |
| L-SSLlabeled training size (Nl)=129732025.08 | 31.8 | |
| SVMK=0.5, Nl=25952025.08 | 31.6 | |
| Logistic RegressionN_l=129732025.08 | 30 | |
| Llabeled training size (Nl)=129732025.08 | 28.3 | |
| KNNK=0.5, Nl=25952025.08 | 25.4 | |
| Semi-Supervised Logistic RegressionN_l=51902025.08 | 25.4 | |
| L-SSLlabeled training size (Nl)=51902025.08 | 24 | |
| K-Nearest NeighborsN_l=51902025.08 | 23.9 | |
| Semi-Supervised Logistic RegressionN_l=38932025.08 | 23.6 | |
| L-SSLlabeled training size (Nl)=38932025.08 | 22.4 | |
| K-Nearest NeighborsN_l=38932025.08 | 21.4 | |
| Logistic RegressionN_l=51902025.08 | 20.9 |