Wolfram Mathematica

« first day (2962 days earlier) ← previous day next day → last day (1512 days later) »

halirutan

3:19 AM

@CarlLange I wouldn't bother too much. You know, the internet will let you when you're wrong :)

The famous KenM

1 hour later…

Carl Lange

4:37 AM

:) OK, thanks.

kirma

5:21 AM

@CATrevillian No problem. :)

1 hour later…

Kuba

6:48 AM

I want to avoid using additional cloud objects but yes, that is a plan B.
Can it be done this way? I do not know but I don't see why not, yet. Delayed FormFunction can capture parent HTTPRequestData so that is ok. The problem is, what happens on submit? I need to 'invoke' the same form using customized HTTPRequest but I am not sure what is the way it should be done.
I even tried `FormFunction[..., URLRead[(*spec to give me html of the parent form for the new request*)]]&` but that never finishes.

1 hour later…

Szabolcs

7:54 AM

I'm sorry, I can't spend time in the chat right now.

Could you please gather whatever information you have and email it to me? When that's done, please also evaluate and send me the troubleshooting notebook. http://szhorvat.net/pelican/files/MaTeX-Troubleshooting.nb

I'll take a look at all of it, probably during the weekend.

Kuba

8:20 AM

Where exactly FormFunction to HTML export is defined? I suppose somewhere around CloudObject` but I am going in circles there.

7 hours later…

JimB

2:50 PM

@kirma The curve you show (Exp[a x^2 + b x + c]) has to eventually shoot straight up or straight down depending on the sign of a. However, prior to that happening that curve seems to fit well. It's the extrapolation that's a problem.

Another curve that many folks are fitting for the recent corona virus is the logistic curve: L/(1+Exp[-k(t-t0)]) where L, k, and t0 are constants to be estimated and t is the day number. That curve by it's very nature has to flatten out. This curve fits very well the currently available data for many countries but the good fit is only because the flattening out hasn't occurred yet. Folks are extrapolating inappropriately again.

Both of those curves fit better than the exponential model but mainly because they both have 3 parameters to play with rather than just 2. (Essentially the model you show is an exponential with the extra squared time value thrown in.)

The lack of accounting for the dependence among observations (because the counts are accumulations) is still not being considered by the many hobbyists making predictions.

kirma

@JimB I don't even try to make actual predictions... :)

5 hours later…