3Blue1Brown epidemic simulations

[EdLee]

Another excellent clip from Grant:

and MinutePhysics:

[Joaz Banbeck]

We must remember that such models are only as good as their data. Certain governments - most notably China's - seem to have fudged their figures a bit.
The only set of sharply decreasing numbers that I would trust is the South Korean's.

[macelee]

Certain governments - most notably China's - seem to have fudged their figures a bit.

You'd better show some proof of this. It is so irresponsible to just make such claims without giving any explanations.

[Mike Novack]

I will give MY explanation for doubt.

Had the Chinese reported the rise in cases ending and then a more or less gradual decline in new cases I would be more inclined to be believing. But a sudden, almost immediate drop to no new cases ??????

Mind that would not be impossible, BUT the scenarios making it possible I do not wish to consider.

[jlt]

It didn't stop immediately in China.

The actual curve is a gradual decline, like at 6:08 of the first video.

[Bill Spight]

I will give MY explanation for doubt.

Had the Chinese reported the rise in cases ending and then a more or less gradual decline in new cases I would be more inclined to be believing. But a sudden, almost immediate drop to no new cases ??????

Mind that would not be impossible, BUT the scenarios making it possible I do not wish to consider.

It's not a sudden, immediate drop. The horizontal dimension is in the log of total cases, not time. Going from, say, 30,000 total cases to 30,500 cases takes up less distance in that dimension than going from 10,000 total cases to 10.500 total cases, even if it takes the same amount of time to see those extra 500 cases. One virtue, if you will, of using the log of total cases instead of time is that the shift from exponential growth is quite obvious and dramatic. OTOH, a slowdown in the rate of growth of cases could still be exponential, like a reduction in the rate of inflation.

[macelee]

The horizontal dimension is in the log of total cases, not time.

This is exactly the reason and this is explained in that youtube video.

[Uberdude]

The reports I've seen about "China reports no new cases" are actually "China reports no new domestically transmitted cases (but 50 reimported ones)", so they've moved the goalposts by not counting imported cases in the headline statistic.

[jlt]

I plotted some graphs for Italy (blue lines) for the last 31 days. Let f(n) be the average number of confirmed cases during the week preceding day n. On the first graph I plotted

x = ln(f(n))
y = ln(f(n)-f(n-1))

i.e. the x-axis represents the total number of cases, smoothed over 1 week, and the y-axis represents the number of new cases, smmothed, and both axes are on a logarithmic scale, like in the second video. It seems that the blue line is starting to deviate from the straight red line.

log-tot-log-new.PNG (11.67 KiB) Viewed 9911 times

On the second graph, I plotted
x = n
y = f(n)-f(n-1)

i.e. I plotted the number of new cases, smoothed, against time. Both axes are on a linear scale. The peak is more visible.

new-cases-smoothed.PNG (23.7 KiB) Viewed 9911 times

So it seems we are seeing the end of the tunnel. Let's hope the trend will be confirmed in the future.

[jann]

So it seems we are seeing the end of the tunnel.

Since most countries started heavy restrictions in the past few weeks, you are seeing the effects of those. This is not the real end of the tunnel (only real stop is vaccine or mass immunity after >50% of pop recovered).

[jlt]

Sure, I am aware of that, but at least we can expect to come back to more or less normal life some time in May, while keeping restrictions like: isolating sick people or people who have been in contact with sick people, avoiding mass gatherings, border controls, widespread use of masks, etc. So my guess is that the European Go Congress will be cancelled.