EpiNow2 and data format

[HAKGH]

Hello,
Can I ask whether EpiNow2 can work on data sets that have weekly or monthly counts Nostradamus of daily counts to minimize calculation time? And how can it be done?
Best regards,
Hanan

[seabbs]

It can work with these counts but internally will run on a daily time-scale and so this won't reduce runtimes.

[HAKGH]

How? what do we have to modify in the code?
Thanks

[seabbs]

You modify week_length to that of your aggregation. Note that this will still estimate a day of the week effect so for long timespans (i.e months) it is likely not very efficient. I am not sure anything useful can really be got from monthly counts though so perhaps this is not an issue. For weekly counts you should essentially be able to use as is though you may need to modifiy the limit for cases corrected when zero (i.e zero_threshold in estimate_infections).

[seabbs]

If your interest is primarily about reducing computation times though a better bet might be to use a different assumption for the gaussian process (see the options for rt_opts) or to use the backcalc/deconvolution method or to fit to a shorter time-series if your interest is in real-time estimates.

[HAKGH]

Thank you