Is the predictive model wrong?

[George K]

Many of us have been following the predictions at this site:

https://covid19.healthdata.org/projections

Sean Davis looked at the predictions and the actuality:

Here's New York:

[Mik]

And this is bad news?

[George K]

@Mik said in Is the predictive model wrong?:

And this is bad news?

So, you're a "half-full" type of guy, LOL?

Apparently, this is the model that is the basis for government action.

[Mik]

All models. See Global Warmingclimate change.

[Mik]

I'd rather have them erring on the side of caution. This model may just be behind by a week or two. Either that or distancing has worked better than anyone thought.

[89th]

I'd have to imagine with most of the country essentially staying at home (really, really, really an unprecedented time we are in!), there's probably a much better impact on the virus than expected. Just think about those "scattering dots" diagrams of how a virus spreads and what if many of the dots are suddenly isolated? Massively good impact. Fingers crossed...

[jon-nyc]

Fingers crossed.

FWIW Cuomo discussed models they look at yesterday. He said the forecast for peak resource use varies from 7 days to 31 days. The IHME model is 7 days (April 9th peak).

We're still WAY under testing as evidenced by our 38% positive test rate. So I wouldn't be surprised if the models showing a later peak are more accurate. I don't know enough about the other states to comment.

Still, wasn't the IHME model supposed to be updated? Why wouldn't they update the curve with actuals?

[jon-nyc]

Two models NY is using besides the IHME.

No labels on y axes but if you eyeball it the numbers for today are in the right spot.

[jon-nyc]

https://13wham.com/news/local/watch-gov-cuomo-holds-daily-covid-19-briefing-from-albany-04-01-2020

[89th]

I created a thread for this but deleted it. I didn't think it could be true.

Gov. Andrew Cuomo said Friday that 2,935 New Yorkers died from the coronavirus over the last 24-hours, the highest single increase in deaths since the outbreak began a few weeks ago.

https://www.cnbc.com/2020/04/03/new-york-gov-cuomo-says-state-saw-its-single-biggest-increase-in-coronavirus-deaths-yesterday.html

[89th]

.........and they've updated the article. The change in bold.

Gov. Andrew Cuomo said Friday that 2,935 New Yorkers have died from the coronavirus so far with 562 new deaths over the last 24 hours

[Copper]

Impeach him!

[jon-nyc]

The more I look into this model the less I like it. It strikes me as more of a ‘technical analysis’ than a ‘fundamental analysis’ in that it’s basically a curve fitting exercise heavily relying on the Wuhan experience.

It seems to be predicting way too early of a peak and is overall too optimistic about our social-distancing measures.

I would love to see the details behind the McKinsey models.

[George K]

@jon-nyc yes.

And the simple fact that its accuracy, in historical perspective, is pretty poor.

[LuFins Dad]

@jon-nyc Which model? The covid19.healthdata.org model? I’ve been calling that model crap for the opposite reason. It undervalues decisions made by the general public and overvalues decisions handed down by the government.

[jon-nyc]

Yes that model. That’s not really the opposite of my view, I also think it overstates the effect of our lockdown.

[LuFins Dad]

@jon-nyc do you have a link to the McKinsey models?

[jon-nyc]

They’re not public to my knowledge.

[George K]

[LuFins Dad]

I’d like to see the ventilator numbers. And Florida is way too early to take anything from a 1 day aberration