After 100 resolved journal questions your calibration table is:
| forecast | observed frequency | |
|---|---|---|
| 0.2 | 30 | 0.30 |
| 0.5 | 40 | 0.45 |
| 0.8 | 30 | 0.70 |
(a) Compute REL, RES, UNC and the Brier score, and check the Murphy identity. (b) Compute the Brier skill score against a coin flip and against climatology. (c) If you had issued the observed frequencies instead (0.3, 0.45, 0.7), what would your Brier score have been? Which component changed? What does this suggest about your forecasting style?