From: Iowan in MD/DC
I agree, CAP. However, this is only a day or two into the model, so you'd expect it to hit close to the mark. On the other hand, I'd expect big perturbations in daily cases. I didn't mean to suggest that 6% under prediction would be a norm, but rather than the daily comparison would be something to keep an eye on.
Hey, that source predicts 93k deaths in the USA. That's alot of people, but way better than so many other predictions, ranging from the famous "100k to 200k" to the earlier "1.7 million."
ORIGINAL: Cap Mandrake
This is one of the three charts in the link given by Erik Rutens yesterday (link: https://covid19.healthdata.org/projections)
The image here shows the graph of expected mortality by day in the USA. For April 2, the projection was 1,036. The actual number (by Worldometers) was 968. So nearly 70 below the prediction, or roughly 6%.
If the daily number continues to fall beneath the projection, the death toll will, of course, be lower than projected.
The total mortality projected by this source is 93,000, with a range of 40k to about 178k. If we come in under the 93k, or significantly under, that's going to be remarkable.
Still, a model within 6% of the actual value is pretty damned good. A Hell of a lot better than the daily Racing Form. It does seem someone fed Governor Cuomo some alarmist ventilator need numbers. Even so, it must be a stressful job. The NY Mayor seems to have become a bit unglued. He's still holding up better than Ray Nagin did.
Worth noting that the previous 1.7M was under the assumption that social distancing restrictions were not put in place, either in time or to a sufficient degree.
If that's the 1.7M from Dr. Ferguson at the Imperial College, he's since revised that downwards based on the change in facts on the ground.
I like the model that Erik linked better for the US, however (and which I've seen in at least 1 other place, as I believe I cited a different link perhaps a day before he posted that one). Not going to deny that part of that preference is I'm going to give less credence to a UK researcher on numbers for the US.