search

sbellan61 / EbolaVaccSim

0 stars 1 forks source link

cluster Oc=6 has highest averted/spent despite being 6th in cumulative risk (supposedly) #11

Closed sbellan61 closed 7 years ago

sbellan61 commented 7 years ago

right now have Oc (original cluster), cluster (from simFuns), and cluster (from ExtractFxns procIrsk) which denote original cluster # before any reordering, reordered for vaccination days, and reordered for plotting clusters in risk-order at the end, respectively.

the latter 2 should be equivalent but seem not to be. for some reason Oc=6 seems to have very high risk spent/averted, when that should follow the track of cumulative risk.

ideas for debugging: rename 3rd cluster variable, rerun equisumm, & see if discrepant w/ 2nd plot vaccdays for each cluster along with it's hazard timeline to see if there's something weird going on. Hard to think of what it could be because still seeing problem for RCT w/o RP. But that suggests that w/o RP might have a bug in them too because they should look very different if this is due to vaccdays. Possible that cluster #ing just got messed up somewhere...

sbellan61 commented 7 years ago

think it's fixed but need to check next by plotting newest output from eqiupsumm

sbellan61 commented 7 years ago

resolved! was just that ordering by cumulative hazard doesn't actually reflect the vaccination ordering so plot looked funky.