search

parksw3 / epidist-paper

Other
10 stars 4 forks source link

Sample size and replicates #22

Closed seabbs closed 2 years ago

seabbs commented 2 years ago

As you suggested I think the current maximum sample size we are investigating is perhaps a bit large given the runtime issues. If we instead drop down to explore n = 10, 100, 200, 400 we have more head room for looking at replicates etc which seems like it will be more useful. As long we discuss the scaling properties of each model (i.e across our explored samples) it feels like we can still make the same point as we would with a larger sample.

seabbs commented 2 years ago

After we chatted I think the decision is to explore

n = 10, 100, 200 for outbreak settings and n = 200 for fixed growth rate settings. Will make this update to the pipeline.

parksw3 commented 2 years ago

Sounds awesome. Thank you! Will do more work on this when I get home.

seabbs commented 2 years ago

Running this now so should be updated tomorrow.