Reshaping ChickWeight to include NA values in dplyr.

[Ven]

BLUF:
How would the following be emulated using dplyr?

• Using the ChickWeight dataset to practice.
• Primarily use data.table
• Linear plots suggest 'missing' chicks.
• Wanted to create NA values in dataset to easily pull missing chicks.

library(data.table)

dt <- data.table(ChickWeight)

# Personal preference and not at all necessary; format variable names to lowercase
dt <- dt[, .(weight, time = Time, chick = Chick, diet = Diet)]

# Reshaping data to include NAs
dt_1 <- melt(
  dcast(dt,
        chick ~ time,
        value.var = c("weight", "diet")),
  id.vars = "chick",
  measure.vars = patterns("^weight", "^diet"),
  variable.name = "time",
  value.name = c("weight", "diet")
)

# NA results
dt_1[is.na(weight)][order(chick, time)]

This was a fun little project and I'd love to learn more.

[JackDavison]

I'm not super familiar with data.table - am I right in saying you're inserting the missing combinations of chick and time, creating NA values for weight and diet? If so, in the tidyverse you don't need to reshape - there's a tidyr function called complete() that does that for you:

library(tidyverse, quietly = T)

tibble(ChickWeight) %>% 
  janitor::clean_names() %>% 
  complete(chick, time) %>% 
  filter(is.na(weight) & is.na(diet))
#> # A tibble: 22 × 4
#>    chick  time weight diet 
#>    <ord> <dbl>  <dbl> <fct>
#>  1 18        4     NA <NA> 
#>  2 18        6     NA <NA> 
#>  3 18        8     NA <NA> 
#>  4 18       10     NA <NA> 
#>  5 18       12     NA <NA> 
#>  6 18       14     NA <NA> 
#>  7 18       16     NA <NA> 
#>  8 18       18     NA <NA> 
#>  9 18       20     NA <NA> 
#> 10 18       21     NA <NA> 
#> # … with 12 more rows

^{Created on 2022-07-14 by the reprex package (v2.0.1)}

I note that, in your example, your "time" column transforms from 0 - 21 in dt to 0 - 12 in dt_1, so perhaps this isn't equivalent.