Groups are the filtered subsets of data which you want to fit statistical models on. This function supports both string inputs and logical functions of variables (provided that the data is already set for the aba model). The inputs should be separated by a comma, where each input is a different group. You can also specify .labels for each group.
Arguments
- .model
An aba model. The model for which you want to set groups.
- ...
comma-separated strings or logical expressions. This specifies the subsets of the data by which the aba model will filter.
- .labels
vector of strings. Optional .labels for printing & plotting.
Details
Note that everyone() or "everyone()" can be used to specify a group with
no filtering. This can be useful when you want to fit models on the entire
group and on a sub-group.
Examples
data <- adnimerge %>% dplyr::filter(VISCODE == 'bl')
# set groups based on logical expressions. Here, data must be supplied first.
model <- data %>% aba_model() %>%
set_groups(
everyone(),
DX_bl == 'CU',
(DX_bl %in% c('MCI','AD')) & (CSF_ABETA_bl < 880)
)
print(model)
#> ----------------------
#> ABA MODEL (not fitted)
#> ----------------------
#> Groups:
#> Everyone
#> DX_bl == "CU"
#> (DX_bl %in% c("MCI", "AD")) & (CSF_ABETA_bl < 880)
# specify .labels which will be used later for printing & plotting
model <- data %>% aba_model() %>%
set_groups(
everyone(),
DX_bl == 'CU',
(DX_bl %in% c('MCI','AD')) & (CSF_ABETA_bl < 880),
.labels = c('All participants', 'CU-only', 'Ab+ MCI & AD')
)
print(model)
#> ----------------------
#> ABA MODEL (not fitted)
#> ----------------------
#> Groups:
#> All participants
#> CU-only
#> Ab+ MCI & AD
# set groups based on strings. No data is required to be supplied first.
model <- aba_model() %>%
set_groups(
"everyone()",
"DX_bl == 'CU'",
"(DX_bl %in% c('MCI','AD')) & (CSF_ABETA_bl < 880)"
)
print(model)
#> ----------------------
#> ABA MODEL (not fitted)
#> ----------------------
#> Groups:
#> Everyone
#> DX_bl == "CU"
#> (DX_bl %in% c("MCI", "AD")) & (CSF_ABETA_bl < 880)