The model did pretty well again: predicted 189, and actuals came in at 164. King County still isn’t reporting negative tests, total tests, or any timeline toward doing so. Weak.
Summary of tomorrow’s predictions:
- Standard model (aka base case): 181
- Lagged model (aka worst case): 205
- Polynomial model (aka best case): 152
I revised the model to take out the 3 most recent days and get a better fit (and potentially better results). We’ll see. For tomorrow, the standard model predicts 181, and, by removing the 3 most recent days due to incomplete counts, the lagged model predicts 205.
One more thought: One model that fits slightly better than a linear model is a polynomial model where there’s a crest and then infections start to decline. It’s slightly more predictive than the linear model. This will be something to watch. In this model, infections are predicted to continue to increase for the next couple of days, then start to decline. The polynomial model predicts 152 tomorrow.