Mark first occurance based on two columns-CodePudding

I have two columns x and y. I want to mark the first occurrence of a value that occurs either in x OR y.

  library(dplyr)
tibble(x = c(1, 2, 2, 3, 7), y = c(7, 7, 8, 9, 10))
#> # A tibble: 5 × 2
#>       x     y
#>   <dbl> <dbl>
#> 1     1     7
#> 2     2     7
#> 3     2     8
#> 4     3     9
#> 5     7    10

^{Created on 2021-11-15 by the reprex package (v0.3.0)}

My expected output should look like:

#> # A tibble: 5 × 4
#>       x     y first_occurance_x first_occurance_y
#>   <dbl> <dbl>             <dbl>             <dbl>
#> 1     1     7                T                T
#> 2     2     7                T                F
#> 3     2     8                F                T
#> 4     3     9                T                T
#> 5     7    10                F                T

^{Created on 2021-11-15 by the reprex package (v0.3.0)}

CodePudding user response：

cbind(df, t(array(!duplicated(c(t(df))), rev(dim(df)))))

  x  y     1     2
1 1  7  TRUE  TRUE
2 2  7  TRUE FALSE
3 2  8 FALSE  TRUE
4 3  9  TRUE  TRUE
5 7 10 FALSE  TRUE

CodePudding user response：

We may pivot to long format and then use duplicated on the 'long' format data and then reshape back to 'wide' format with pivot_wider

library(dplyr)
library(tidyr)
library(stringr)
df1 <- df1 %>% 
      mutate(rn = row_number()) %>%
      pivot_longer(cols = c(x, y)) %>% 
      mutate(value = !duplicated(value),
           name = str_c('first_occurance_', name)) %>% 
      pivot_wider(names_from = name, values_from = value) %>% 
      select(-rn) %>% 
      bind_cols(df1, .)

-output

df1
# A tibble: 5 × 4
      x     y first_occurance_x first_occurance_y
  <dbl> <dbl> <lgl>             <lgl>            
1     1     7 TRUE              TRUE             
2     2     7 TRUE              FALSE            
3     2     8 FALSE             TRUE             
4     3     9 TRUE              TRUE             
5     7    10 FALSE             TRUE

data

df1 <- tibble(x = c(1, 2, 2, 3, 7), y = c(7, 7, 8, 9, 10))

CodePudding user response：

A possible solution:

library(tidyverse)

df <- tibble(x = c(1, 2, 2, 3, 7), y = c(7, 7, 8, 9, 10))

df %>% 
  group_by(x) %>% 
  mutate(firstx = ifelse(row_number() == 1, T, F)) %>% 
  ungroup %>% 
  group_by(y) %>% 
  mutate(firsty = ifelse(row_number() == 1, T, F)) %>% 
  ungroup %>% 
  mutate(
    aux1 = x %in% lead(y), aux2 = y %in% lead(x),
    firstx = ifelse(aux1 & row_number() != 1, F, firstx),
    firsty = ifelse(aux2 & row_number() != 1, F, firsty),
    aux1 = NULL, aux2 = NULL)

#> # A tibble: 5 × 4
#>       x     y firstx firsty
#>   <dbl> <dbl> <lgl>  <lgl> 
#> 1     1     7 TRUE   TRUE  
#> 2     2     7 TRUE   FALSE 
#> 3     2     8 FALSE  TRUE  
#> 4     3     9 TRUE   TRUE  
#> 5     7    10 FALSE  TRUE