After completion of intial PR https://github.com/epinowcast/epidist/pull/125 adding as_latent_individual functionality we should continue to build out the range of data which can be processed and run with the latent_individual model. To enable date format data to be processed I propose to add a helper function mapping it to a format which can be accepted by existing as_latent_individual functionality. More specifically:
A. General format (as accepted by current as_latent_individual):
column
type
case
int
ptime_lwr
num
ptime_upr
num
stime_lwr
num
stime_upr
num
obs_at
num
B. Only dates format:
column
type
case
int
ptime_date
num or date
stime_date
num or date
obs_at
num
Proposed workflow would be:
A <- function_in_this_issue(B)
prep_obs <- as_latent_individual(A)
Maybe or out of scope for this issue
A date is just one example of a fixed censoring window. Can we have other fixed censoring windows work with this function? I think should be possible but would call this "nice-to-have"
Things to do with models outside latent_individual. Maybe this is a mistake and need to think more about how useful moving to type A will be in general?
Uncertainties
Could argue about this approach versus prep_obs <- function_in_this_issue(B) direct routes? i.e. putting the transformation inside as_latent_individual.
After completion of intial PR https://github.com/epinowcast/epidist/pull/125 adding
as_latent_individual
functionality we should continue to build out the range of data which can be processed and run with thelatent_individual
model. To enable date format data to be processed I propose to add a helper function mapping it to a format which can be accepted by existingas_latent_individual
functionality. More specifically:A. General format (as accepted by current
as_latent_individual
):B. Only dates format:
Proposed workflow would be:
Maybe or out of scope for this issue
latent_individual
. Maybe this is a mistake and need to think more about how useful moving to type A will be in general?Uncertainties
Could argue about this approach versus
prep_obs <- function_in_this_issue(B)
direct routes? i.e. putting the transformation insideas_latent_individual
.New issues