Reading the reports¶
Under testing
Report wording and the trust rating are under active testing. Every sentence is generated by fixed rules from the run's own numbers (no AI-generated text), but the rules themselves are still being refined.
Both v2 reports (forecast and climate scenarios) start with a plain-language section and end with a collapsible Technical details section for specialists.
Forecast report¶
| Section | What it tells you |
|---|---|
| The short version | Expected total, busiest month, comparison with a typical year, which months look higher or lower than usual, and how many already-known months fell inside the likely range |
| Expected cases, month by month | A chart and table: expected cases, likely range (8 in 10 chance), comparison with a typical year and, where known, what actually happened |
| How much can you trust this? | A Good / Moderate / Low rating with its reasons |
| Things to keep in mind | Caveats that apply to this run |
| Technical details | Every model, every score, validation rules and the C-DSI v2 interpretation |
The trust rating¶
One point each for: (1) the likely range contained the real number about as often as it should; (2) the method shown beat the "same as recent years" baseline in past tests; (3) the methods broadly agree; (4) the forecast looks no more than 12 months ahead. 4 points = Good, 2–3 = Moderate, 0–1 = Low. "Not tested" means there was no past data to check against.
Note
A "Good" rating means the method did well in past tests on this dataset. It is not a guarantee.
Scenario report¶
Plain sentences per emissions pathway (cases a year, change from the baseline, likely range), a confidence rating (backtest, climate-model agreement, sensitivity, and how clearly the data identify which climate factors matter), and what the projections leave out. Decade tables, the backtest, the tree-model comparison and methods are in the technical details.
Words used¶
Likely range: where the real number should fall 8 times out of 10. Typical year: average for the same month over the last five years. Past tests / hindcasts: forecasts made from earlier dates and checked against what happened. Ensemble: the middle value of all models. Emissions pathway (SSP): a standard scenario for future greenhouse-gas emissions.