Performance of the ensemble random forests model of return success and time at large measured by three threshold-independent metrics: area under the curve (AUC), root mean squared error (RMSE), and the true skill statistic (TSS). There were four possible states in the model, no returns, returning after 2 years at large, returning after 3 years at large, and returning after 4 years at large, each with their own performance.
State . | AUC . | RMSE . | TSS . |
---|---|---|---|
No returns | 0.89 | 0.47 | 0.62 |
2 years at large | 0.99 | 0.21 | 0.94 |
3 years at large | 0.9 | 0.35 | 0.63 |
4 years at large | 1 | 0.21 | 0.98 |
State . | AUC . | RMSE . | TSS . |
---|---|---|---|
No returns | 0.89 | 0.47 | 0.62 |
2 years at large | 0.99 | 0.21 | 0.94 |
3 years at large | 0.9 | 0.35 | 0.63 |
4 years at large | 1 | 0.21 | 0.98 |
Performance of the ensemble random forests model of return success and time at large measured by three threshold-independent metrics: area under the curve (AUC), root mean squared error (RMSE), and the true skill statistic (TSS). There were four possible states in the model, no returns, returning after 2 years at large, returning after 3 years at large, and returning after 4 years at large, each with their own performance.
State . | AUC . | RMSE . | TSS . |
---|---|---|---|
No returns | 0.89 | 0.47 | 0.62 |
2 years at large | 0.99 | 0.21 | 0.94 |
3 years at large | 0.9 | 0.35 | 0.63 |
4 years at large | 1 | 0.21 | 0.98 |
State . | AUC . | RMSE . | TSS . |
---|---|---|---|
No returns | 0.89 | 0.47 | 0.62 |
2 years at large | 0.99 | 0.21 | 0.94 |
3 years at large | 0.9 | 0.35 | 0.63 |
4 years at large | 1 | 0.21 | 0.98 |
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.