BUGS
- if data has no injuries or all injuries and is processed with
external.sep=TRUE
, if fails with a cryptic indexing error - if you pass a burnin greater than the Nsim you get a cryptic java error
- if you redefine the conditional probability and no causes+symptoms pairs are
impossible and you do not pass
exclude.impossible.cause=FALSE
, it fails with a cryptic data is null error. This happens when you go to castimpossible <- as.matrix(impossible)
(InSilicoVA/R/insilico_core.r line 628) - Doesn't fully validate 'type' when calling
extract.prob
. This leads to an indexing error later.
API
nlevels.dev
has a hidden default at InSilicoVA/R/insilico_core.r lines 509 to 514
UX
data
attribute on returned insilico object only has dimnames for rows. colnames have been dropped.data
andindiv.prob
on returned insilico object have different number of observations.
DEV API
- using warning() vs cat() for printing warnings warning InSilicoVA/R/insilico_core.r line 548
- No way to specify dev external causes. InSilicoVA/R/insilico_core.r line 628
sets
external.causes <- NULL
- if you pass
customization.dev=TRUE
,isNumeric=TRUE
,external.sep=FALSE
, anddatacheck=FALSE
are numerics converted to strings when looking for missing InSilicoVA/R/insilico_core.r line 663