Regret Minimization on Discrete feature values setting n=5000 (simulation)
0.17Mean RegretSUPER
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SUPERepsilon=0.9, Sample size (n)=5000, Replications=502022.09 | 0.17 | 0.086 | |
| SUPERepsilon=0.7, Sample size (n)=5000, Replications=502022.09 | 0.18 | 0.035 | |
| SZONLYepsilon=0.5, Sample size (n)=5000, Replications=502022.09 | 0.21 | 0.017 | |
| SUPERepsilon=0.5, Sample size (n)=5000, Replications=502022.09 | 0.21 | 0.014 | |
| SZONLYepsilon=0.7, Sample size (n)=5000, Replications=502022.09 | 0.22 | 0.018 | |
| SZONLYepsilon=0.9, Sample size (n)=5000, Replications=502022.09 | 0.24 | 0.012 | |
| SONLYepsilon=0.5, Sample size (n)=5000, Replications=502022.09 | 0.25 | 0.0003 | |
| SONLYepsilon=0.7, Sample size (n)=5000, Replications=502022.09 | 0.25 | 0.0003 | |
| SONLYepsilon=0.9, Sample size (n)=5000, Replications=502022.09 | 0.25 | 0.0003 |