Extract values based on conidtion of another column

Question

case_number <- c(\"1\", \"1\", \"2\", \"2\", \"2\", \"3\", \"3\") 
type <- c(\"STD\", \"STD2\", \"STD\", \"STD3\", \"STD2\", \"STD\", \"STD2\") 
date <- as.Date         


        

Answer 1

Assuming every case_number would have both the values , another option is to check the position of "STD" and "STD2" and select groups where the difference is equal to 1.

check_fun <- function(x) {
   abs(diff(c(which.max(x == "STD"), which.max(x == "STD2")))) == 1
}

library(dplyr)

data %>% group_by(case_number) %>% filter(check_fun(type))

# case_number type  date      
#  <fct>       <fct> <date>    
#1 1           STD   2008-11-01
#2 1           STD2  2009-03-25
#3 3           STD   2015-03-14
#4 3           STD2  2015-04-15

Or if you just need the unique case_number

data %>% 
  group_by(case_number) %>% 
  filter(check_fun(type)) %>% 
  pull(case_number) %>% 
  unique

#[1] 1 3
#Levels: 1 2 3

Answer 2

Here's a shot:

myfilter <- function(x) {
  r <- rle(x %in% c("STD", "STD2"))
  any(r$lengths[r$values] > 1)
}

library(dplyr)
data %>%
  group_by(case_number) %>%
  filter(myfilter(type)) %>%
  ungroup()
# # A tibble: 4 x 3
#   case_number type  date      
#   <fct>       <fct> <date>    
# 1 1           STD   2008-11-01
# 2 1           STD2  2009-03-25
# 3 3           STD   2015-03-14
# 4 3           STD2  2015-04-15

It does not care about order, just about finding either of them in a chain of two (or more).