New function ppc_km_overlay_grouped() #260
Conversation
|
The failing R-CMD-check on Windows might arise from this "withr" issue. |
Codecov Report
@@ Coverage Diff @@
## master #260 +/- ##
=======================================
Coverage 98.36% 98.37%
=======================================
Files 32 32
Lines 4155 4175 +20
=======================================
+ Hits 4087 4107 +20
Misses 68 68
Continue to review full report at Codecov.
|
… "ppc-censoring" to their own folder).
…shift due to commit df58a47).
fweber144
force-pushed
the
ppc_censoring_grouped
branch
from
1a22a50
to
479fded
Compare
|
Hi @fweber144 thank you for this and really sorry I forgot about reviewing this for so long! I finally have some time to put into bayesplot and I noticed that I had missed this and other PRs when I was super busy with other stuff.
I agree this is a bit hacky but I understand why you went with this option and I'm not sure that I have a better idea. So let's go with this for now and we can always update it later if you or anyone else has an idea for how to improve it. |
Codecov Report
@@ Coverage Diff @@
## master #260 +/- ##
==========================================
- Coverage 98.38% 98.37% -0.01%
==========================================
Files 32 32
Lines 4199 4184 -15
==========================================
- Hits 4131 4116 -15
Misses 68 68
Continue to review full report at Codecov.
|
|
No problem, you must have a lot of work with all those R packages. Thanks for merging! |
I added
ppc_km_overlay_grouped(). Unfortunately, the+ list(data)trick which is used e.g. inppc_ecdf_overlay_grouped()cannot be used inppc_km_overlay_grouped()since if it was used,survival::survfit()would not "know" of the grouping and would not be able to take it into account. The solution I have implemented now is kind of hacky. In particular, it requires to pass the grouping toppc_km_overlay()which I did via the ellipsis (...). I couldn't think of a better solution, but if you have an idea, please let me know.I also performed some minor fixes, e.g. correcting a typo in "args-groups.R" and updating the expected results for the "cmdstanr methods work" unit test. Just have a look at my commits and decide which ones you want to use.