Account-level bot detection on Twitter Bot Detection account-level (user)
100PrecisionAdaBoost Classifier
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AdaBoost ClassifierOversampling technique=SMOTENN2018.02 | 100 | 100 | 100 | 99.81 | 99.81 | |
| Logistic RegressionOversampling technique=SMOTENN2018.02 | 99 | 99 | 99 | 98.59 | 98.62 | |
| Random Forest ClassifierOversampling technique=SMOTENN2018.02 | 99 | 99 | 99 | 99.37 | 99.38 | |
| 2-layer NNOversampling technique=SMOTENN, Architecture=(300,200,1) ReLU+Adam2018.02 | 99 | 99 | 99 | 98.78 | 98.79 | |
| Random Forest ClassifierOversampling technique=SMOTOMEK2018.02 | 99 | 99 | 99 | 98.59 | 98.59 | |
| AdaBoost ClassifierOversampling technique=SMOTOMEK2018.02 | 99 | 99 | 99 | 98.65 | 98.65 | |
| Random Forest ClassifierOversampling technique=None2018.02 | 98 | 98 | 98 | 98.39 | 98.45 | |
| AdaBoost ClassifierOversampling technique=None2018.02 | 98 | 98 | 98 | 98.23 | 98.23 | |
| 2-layer NNOversampling technique=None, Architecture=(500,200,1) ReLU+Adam2018.02 | 95 | 95 | 95 | 94.96 | 94.75 | |
| SGD ClassifierOversampling technique=SMOTENN2018.02 | 95 | 94 | 94 | 94.33 | 94.43 | |
| 2-layer NNOversampling technique=SMOTOMEK, Architecture=(300,200,1) ReLU+Adam2018.02 | 95 | 95 | 95 | 93.91 | 94.89 | |
| Logistic RegressionOversampling technique=None2018.02 | 94 | 93 | 93 | 90.66 | 88.91 | |
| Logistic RegressionOversampling technique=SMOTOMEK2018.02 | 92 | 91 | 91 | 90.94 | 90.98 | |
| SGD ClassifierOversampling technique=SMOTOMEK2018.02 | 90 | 90 | 90 | 90.3 | 90.31 | |
| SGD ClassifierOversampling technique=None2018.02 | 87 | 87 | 87 | 87.26 | 86.8 |