R - pivot_wider - with multiple duplicate columns to pivot

I have a dataframe like the following:

id = c(1,2,3,4,5)
value = c(100, 200, 300, 400, 500)
tech = c('A','B','C','D','E')
tech2 = c(NA, NA,'A','B', NA)

data = data.frame(id, value, tech, tech2)

I want to convert data to the following:

Where tech is only one column, and any id with no NA in tech 2 is duplicated and has its value split by 2 for each tech e.g. id number 3 has two techs, and a value of 300, so each tech gets 150.

I have looked at pivot_wider and pivot_longer but the examples either have numerical values within the tech column, or only one tech column.

Any suggestions?

CodePudding user response：

A possible solution:

library(tidyverse)

data %>% 
  mutate(
    value = if_else(rowSums(is.na(across(tech:tech2))) == 0, value/2, value),
    tech = paste(tech, tech2), tech2 = NULL) %>% 
  separate_rows(tech) %>% 
  filter(tech != "NA")

#> # A tibble: 7 × 3
#>      id value tech 
#>   <dbl> <dbl> <chr>
#> 1     1   100 A    
#> 2     2   200 B    
#> 3     3   150 C    
#> 4     3   150 A    
#> 5     4   200 D    
#> 6     4   200 B    
#> 7     5   500 E

CodePudding user response：

A base R option using reshape ave

transform(
  na.omit(
    reshape(
      data,
      direction = "long",
      idvar = c("id", "value"),
      varying = -c(1:2),
      v.names = "tech"
    )
  )[-3],
  value = value / ave(value, id, FUN = length)
)

gives

        id value tech
1.100.1  1   100    A
2.200.1  2   200    B
3.300.1  3   150    C
4.400.1  4   200    D
5.500.1  5   500    E
3.300.2  3   150    A
4.400.2  4   200    B