Dynamic statements in a case_when function

user_12127172 0
Viewed 184

I have the following data:

test <- tibble(id = 1:4, name = letters[1:4], revenue = seq(100, 400, 100))

From this, I create lagged variables:

 test_lagged <- 
    test %>% 
      mutate(
        revenue_lag_1 = lag(revenue, 1),
        revenue_lag_2 = lag(revenue, 2),
        revenue_lag_3 = lag(revenue, 3),
        across(
          contains("revenue"),
          ~replace_na(., 0)
        )
      )

I already have an approach that helps me do the above step dynamically, so that I can specify the number om lags and lagged variables in a function. However, the next step, I have only managed to implement explicitly (by mapping each column explicitly):

test_lagged %>% 
  mutate(
    any_previous_zero_revenue = case_when(
      
      revenue_lag_1 == 0 |
      revenue_lag_2 == 0 |
      revenue_lag_3 == 0 ~ TRUE,
      
      TRUE ~ FALSE
    )
  )

What I want is to be able to apply this dynamically in a case_when function by using something like "any_of", "all_of", contains(), starts_with() or similar.

First, I want to say that for all columns that contains/starts with "revenue_", check if "any_of" these variables (for a given line) == 0. That way I can create dynamically as many lagged variables as I want to without having to change the case_when function.

dplyr case r contains
2 Answers

Here a solution

library(tidyverse)

test_lagged %>% 
  mutate(
    any_previous_zero_revenue = if_else(if_any(.cols = starts_with("revenue_lag"),`==`,0),TRUE,FALSE)
  )

# A tibble: 4 x 7
     id name  revenue revenue_lag_1 revenue_lag_2 revenue_lag_3 any_previous_zero_revenue 
  <int> <chr>   <dbl>         <dbl>         <dbl>         <dbl> <lgl>
1     1 a         100             0             0             0 TRUE 
2     2 b         200           100             0             0 TRUE 
3     3 c         300           200           100             0 TRUE 
4     4 d         400           300           200           100 FALSE
user_96960370

Using rowSums -

library(dplyr)

test_lagged %>%
  mutate(previous_zero = rowSums(select(., starts_with('revenue_lag')) == 0) > 0)

#     id name  revenue revenue_lag_1 revenue_lag_2 revenue_lag_3 previous_zero
#  <int> <chr>   <dbl>         <dbl>         <dbl>         <dbl> <lgl>        
#1     1 a         100             0             0             0 TRUE         
#2     2 b         200           100             0             0 TRUE         
#3     3 c         300           200           100             0 TRUE         
#4     4 d         400           300           200           100 FALSE
user_39629140
Related