Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
It wasn't that it predicted millions of deaths - though many models did - it was that it (entirely reasonably) predicted we'd have more than 300 thousand people that needed a hospital bed.
That doesn't seem like a lot - does it? When we're looking at Italy / Spain / France (at the time)? The policy makers incorrectly thought folks would care about dying - themselves or others - but as it turns out, lots of people care about the economy more.
If they had emphasized from the beginning the lack of resources we have here (and not just here, to be fair) - it MAY have at least seemed a reasonable conclusion: If we don't do SOMETHING - it's a pretty easy prediction that a few million folks are going to catch this AT THE SAME TIME - and we won't have any place to put them. Not die. Just catch it. Many will die, certainly. Dead people don't need a hospital bed.
Here were are two months later - and several million, if not tens of millions - do indeed have it. Did they all catch it at once? Nope. So it all worked out. Now you have to prove the lockdown had no effect on the rate of transmission - and that it was, instead, magic dust or a powerful spell, or maybe a leprechaun. Because that's all ya got to explain the slow down EXCEPT the "stay at home" orders.
A slow down in infections was the goal. It worked. It's past now. Stop yelling at people for not wanting to get sick. The flu sucks. I do not want it. COVID is worse. I don't want it either. Go live your life, and let other people do what they do.
Cannot rep you again yet ...
Quote:
Originally Posted by Mircea
Unfortunately, no one knew that, because a handful of people with flu-like symptoms is not cause for alarm.
China's Flu Season runs concurrent with the US, October-March.
People dying of flu-like symptoms in October in China or the US is not a shocker.
3,000-5,000/month dying of Flu in the US is not a big deal. China's population is 5x larger, so I'd imagine 10,000-15,000/month dying of Flu is not a big deal there, either.
It's only when you start having more people than normal that anyone starts paying attention.
It was not the number of sick people that sent up a red flag in Wuhan, it was the atypical presentation suggestive of the original SARS virus.
Do you not understand what "projection" means? How about "prediction"?
You know why the model wasn't correct? Because we didn't do nothing; we did something . Good grief ...
Apparently we don't agree with the definition of 'correct'. And please don't lecture me on projection or prediction, if you did have a firm grasp yourself you'd understand that correctness for a model is related to the conditions it was modeling and that is precisely what I was referring to. How do you 'prove' that if the conditions of the model were actually met, that the results of the model would have matched the results in reality. As I quite explicitly stated, you can't from an absolute standpoint because those conditions ended up not matching what happened in reality. What's so hard to grasp about that concept? However, just because the conditions changed (and quite honestly, they almost always do), that doesn't mean that you can't postmortem the model to get a better feel for how it was tracking against reality, and that's key, because no matter what, you need to have some type of quality indicator for those models, you can't just shrug your shoulders and say 'stuff changed, let's just assume the model was correct'. This would not be acceptable for reputable epidemiologists or data scientists.
Like it or not, head slaps and declarations of 'we did something' don't cut it for me, in large part because of what I do in the real world. Model correctness, what it means and how you measure it are not big block concepts for some.
Still doesn't mean the model was correct, unfortunately that can't be proven one way or the other, so some will take it as gospel, some will treat it as a random guess.
The estimate was 2MM is we did nothing, 100k - 200k if we shutdown. We shut down and the prediction is accurate. We'll never know for sure if the 2MM prediction was accurate (thankfully).
Yes and Apparently it’s now gone from “flattening the curve” to until we have a vaccine”
See, that's the beauty of the scientific method: the ability to adapt the method to conditions as they change and as you learn more about the topic in question.
Quote:
Originally Posted by austinnerd
How do you 'prove' that if the conditions of the model were actually met, that the results of the model would have matched the results in reality.
Predict
verb
: to declare or indicate in advance
especially : foretell on the basis of observation, experience, or scientific reason
See, that's the beauty of the scientific method: the ability to adapt the method to conditions as they change and as you learn more about the topic in question.
Predict
verb
: to declare or indicate in advance
especially : foretell on the basis of observation, experience, or scientific reason
Why thank you, all these years of working on predictive models and I never knew what the definition of 'predict' was.
Just as an FYI, from experience 'scientific method' <> 'correctness'. The conditions are just a set of variables in the model, the model creator has other variables and then functions to apply to those variables. Just as the external variables changing can cause the output of the model to change, both the modeler determined variables and the calculations of the functions have an effect on the output. My question, that you have still not answered or even addressed, is how to determine that the internal variables and functions of the models are/were correct. If a data scientist came to me and I asked them how they validated their model and their response was 'using the scientific method' or 'scientific reasoning', they wouldn't have a job for long.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.