`where` in `group_count()` generates error
tdwils opened this issue · comments
Prerequisites
Background: For tables that require more than one section of group counts, a separate filter needs to be applied for each group count. For example, an adverse events summary table shows the number of AEs that are related (AEREL != "NONE"), serious (AESER == "Y"), etc. Each group_count() needs its own filter.
Description
After updating {Tplyr} from v0.4.4 to v1.0.1, the behavior of the where option seems to have changed. When the target dataset is different from the population dataset, and the where condition is placed in the group_count() function, an error is now produced by get_numeric_data() and build().
If the target and population datasets are the same, or if the where is placed in the tplyr_table() function, then it runs without error.
Steps to Reproduce
Reproducible example:
target dataset = adae
population dataset = adsl
filter for this layer: AEREL != "NONE"
library(dplyr)
library(Tplyr)
library(admiral)
library(admiral.test)
data("admiral_adsl")
data("admiral_ae")
adsl <- admiral_adsl
# Basic ADAE dataset
adae <- admiral_ae %>% left_join(adsl, by = c("STUDYID", "USUBJID"))
t <- tplyr_table(adae, TRT01A) %>%
set_pop_data(adsl) %>%
set_pop_treat_var(TRT01A) %>%
set_pop_where(TRUE) %>%
add_layer(
group_count(AEDECOD, where = AEREL != "NONE") %>%
set_distinct_by(USUBJID)
)
num_data <- get_numeric_data(t)
# Error in `filter()`:
# ! Problem while computing `..1 = AEREL != "NONE"`.
# Caused by error:
# ! object 'AEREL' not found
t_df <- build(t)
# Error in `filter()`:
# ! Problem while computing `..1 = AEREL != "NONE"`.
# Caused by error:
# ! object 'AEREL' not found
Expected behavior:
get_numeric_data(t) and build(t) run without error.
Actual behavior:
# Error in `filter()`:
# ! Problem while computing `..1 = AEREL != "NONE"`.
# Caused by error:
# ! object 'AEREL' not found
Versions
R version 4.1.2 (2021-11-01)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19044)
Matrix products: default
locale:
[1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C
[5] LC_TIME=English_United States.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] admiral.test_0.2.0 admiral_0.8.3 Tplyr_1.0.1 dplyr_1.0.8
loaded via a namespace (and not attached):
[1] rstudioapi_0.13 magrittr_2.0.2 hms_1.1.1 insight_0.15.0 tidyselect_1.1.1
[6] R6_2.5.1 rlang_1.0.1 fansi_1.0.2 stringr_1.4.0 tools_4.1.2
[11] utf8_1.2.2 cli_3.1.1 ellipsis_0.3.2 assertthat_0.2.1 tibble_3.1.6
[16] lifecycle_1.0.1 crayon_1.5.0 purrr_0.3.4 tidyr_1.2.0 vctrs_0.3.8
[21] sjlabelled_1.1.8 glue_1.6.1 haven_2.4.3 admiraldev_0.1.0 stringi_1.7.6
[26] compiler_4.1.2 pillar_1.7.0 forcats_0.5.1 generics_0.1.2 lubridate_1.8.0
[31] pkgconfig_2.0.3
Let me know if you need any other information. Thank you!
+1
Agreed. This error has been a roadblock for our team @ Biogen as of {Tplyr} v1.0.1. In the meantime, we are requiring our users to use v0.4.4.
Thanks for the report! I've verified this and we're looking into it!
@elimillera error comes from here
Line 670 in d84e065
The issue is that denom_where is AEREL != "None", so the layer level where is overwriting the population where - which given the pop data, that shouldn't be the case.
Also @tdwils kudos on a really well written bug report!
Thanks @tdwils and @AARON-CLARK. The branch 'bug-fix/issue-106' should have this resolved if you are able to give it a check.
@elimillera Thank you for the quick response! I have confirmed that this issue has been resolved.
Do you have an estimate for when this would be pushed to CRAN?
@tdwils we'll definitely push it up by early next week - I have some comments to Eli to look through but we'll get it pushed back up ASAP.
@tdwils and @AARON-CLARK On it's way to CRAN!
Thank you!