Parameters of the distribution of generation time

[MatsuuraKentaro]

In MLE version, I found:

## generation interval (Nishiura, et al, 2020)
gi_fit = list(shape=2.345, scale=5.452)

On the other hand, in Stan version, I found:

  ## Serial interval [Nishiura et al 2020 - only certain cases]
  param1_SI = 2.305,
  param2_SI = 5.452,

Which is correct, MLE version or Stan version?

How can I confirm this? I couldn't be sure because the following paper doesn't give the numbers in details: https://www.sciencedirect.com/science/article/pii/S1201971220301193

[aakhmetz]

Hello – because generation interval is assumed to be the well approximated by the serial interval, they are the same. Both are modeled by Weibull distribution with the mean 4.85 days and standard deviation 2.30 days (this would give the shape and scale equal 2.345 and 5.452, respectively).

You may check the parameters of SI with the best-fit model with right truncation here (which is "only certain pairs" https://github.com/aakhmetz/nCoVSerialInterval2020)

[MatsuuraKentaro]

Thank you very much for your reply. I can see that the mean of param1 in Out[13] is 2.305 in your link. So, 2.345 in MLE version should be 2.305, right?

[aakhmetz]

Yes, thank you very much for noticing

On Fri, 15 May 2020 at 23:53, Kentaro Matsuura notifications@github.com wrote:

Thank you very much for your reply. I can see that the mean of param1 in Out[13] is 2.305 in your link. So, 2.345 in MLE version should be 2.305, right?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/contactmodel/COVID19-Japan-Reff/issues/3#issuecomment-629278875, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAQSSFSKXCTIHJPZMIJUTM3RRVJNJANCNFSM4NBT62YA .