univariate_associations.RdEvaluate a list of scalar functions on any number of "response" columns by any number of "predictor" columns
univariate_associations(
data,
f,
responses,
predictors
)A data.frame.
A function or a list of functions (preferably named) that take a vector as input in the first two arguments and return a scalar.
A vector of quoted/unquoted columns, positions, and/or tidyselect::select_helpers to be evaluated as the first argument. See the left argument in dish.
A vector of quoted/unquoted columns, positions, and/or tidyselect::select_helpers to be evaluated as the second argument. See the right argument in dish.
A tibble::tibble with the response/predictor columns down the rows and the results of the f across the columns. The names of the result columns will be the names provided in f.
#Make a list of functions to evaluate
f <-
list(
#Compute a univariate p-value
`P-value` =
function(y, x) {
if(some_type(x, c("factor", "character"))) {
p <- fisher.test(factor(y), factor(x), simulate.p.value = TRUE)$p.value
} else {
p <- kruskal.test(x, factor(y))$p.value
}
ifelse(p < 0.001, "<0.001", as.character(round(p, 2)))
},
#Compute difference in AIC model between null model and one predictor model
`AIC Difference` =
function(y, x) {
glm(factor(y)~1, family = "binomial")$aic -
glm(factor(y)~x, family = "binomial")$aic
}
)
#Choose a couple binary outcomes
heart_disease %>%
univariate_associations(
f = f,
responses = c(ExerciseInducedAngina, HeartDisease)
)
#> # A tibble: 14 × 4
#> response predictor `P-value` `AIC Difference`
#> <chr> <chr> <chr> <dbl>
#> 1 ExerciseInducedAngina Age 0.14 0.567
#> 2 ExerciseInducedAngina Sex 0.01 4.72
#> 3 ExerciseInducedAngina ChestPain <0.001 64.1
#> 4 ExerciseInducedAngina BP 0.38 -0.740
#> 5 ExerciseInducedAngina Cholesterol 0.14 -0.875
#> 6 ExerciseInducedAngina BloodSugar 0.66 -1.80
#> 7 ExerciseInducedAngina MaximumHR <0.001 42.5
#> 8 HeartDisease Age <0.001 13.4
#> 9 HeartDisease Sex <0.001 22.0
#> 10 HeartDisease ChestPain <0.001 80.1
#> 11 HeartDisease BP 0.03 4.95
#> 12 HeartDisease Cholesterol 0.04 0.206
#> 13 HeartDisease BloodSugar 0.66 -1.81
#> 14 HeartDisease MaximumHR <0.001 55.1
#Use a subset of predictors
heart_disease %>%
univariate_associations(
f = f,
responses = c(ExerciseInducedAngina, HeartDisease),
predictors = c(Age, BP)
)
#> # A tibble: 4 × 4
#> response predictor `P-value` `AIC Difference`
#> <chr> <chr> <chr> <dbl>
#> 1 ExerciseInducedAngina Age 0.14 0.567
#> 2 ExerciseInducedAngina BP 0.38 -0.740
#> 3 HeartDisease Age <0.001 13.4
#> 4 HeartDisease BP 0.03 4.95
#Numeric predictors only
heart_disease %>%
univariate_associations(
f = f,
responses = c(ExerciseInducedAngina, HeartDisease),
predictors = is.numeric
)
#> Warning: Use of bare predicate functions was deprecated in tidyselect 1.1.0.
#> ℹ Please use wrap predicates in `where()` instead.
#> # Was:
#> data %>% select(is.numeric)
#>
#> # Now:
#> data %>% select(where(is.numeric))
#> # A tibble: 8 × 4
#> response predictor `P-value` `AIC Difference`
#> <chr> <chr> <chr> <dbl>
#> 1 ExerciseInducedAngina Age 0.14 0.567
#> 2 ExerciseInducedAngina BP 0.38 -0.740
#> 3 ExerciseInducedAngina Cholesterol 0.14 -0.875
#> 4 ExerciseInducedAngina MaximumHR <0.001 42.5
#> 5 HeartDisease Age <0.001 13.4
#> 6 HeartDisease BP 0.03 4.95
#> 7 HeartDisease Cholesterol 0.04 0.206
#> 8 HeartDisease MaximumHR <0.001 55.1