I have a dataframe in R where I need to shift the values up in a particular column using a numeric value. The numeric value which is used as input to shift the values up are associated with a grouping variable and are often different from group to group. Here is an example of a mock dataframe

mock data frame
df<-data.frame(ID=c("a","a","a","a","a","b","b","b","b","b","c","c","c","c","c"),
               var_description = c("description 1", "description 2","description 3","description 4","description 5",
                                   "description 1", "description 2","description 3","description 4","description 5",
                                   "description 1", "description 2","description 3","description 4","description 5"),
               var_value = c(NA,NA,NA,"desc 1 val","desc 2 val",
                             NA,"desc 1 val","desc 2 val","desc 3 val","desc 4 val",
                             NA,NA,NA,NA,"desc 1 val"),
               shift_val = c(3,3,3,3,3,
                             1,1,1,1,1,
                             4,4,4,4,4))

Here is the current working solution:-

#getting unique vals
unique_id = unique(df$ID)

#initialising dataframe to collect results
df_results<-data.frame()


#function to shift values up
shift <- function(x,n){
  c(x[-(seq(n))],rep(NA,n))
}


#for loop

for (i in unique_id) {
  
  #filtering by each ID
  one_id<-df%>%filter(ID==i)
  
  #getting value to shift values by
  Move_by_val = unique(one_id$shift_val)
  #shifting
  one_id$new_var_value<-shift(one_id$var_value, Move_by_val)
  
  #binding one_id onto df_results
  
  df_results<-rbind(df_results,one_id)
  
}


head(df_results,10)

# ID var_description  var_value shift_val new_var_value
# 1   a   description 1       <NA>         3    desc 1 val
# 2   a   description 2       <NA>         3    desc 2 val
# 3   a   description 3       <NA>         3          <NA>
# 4   a   description 4 desc 1 val         3          <NA>
# 5   a   description 5 desc 2 val         3          <NA>
# 6   b   description 1       <NA>         1    desc 1 val
# 7   b   description 2 desc 1 val         1    desc 2 val
# 8   b   description 3 desc 2 val         1    desc 3 val
# 9   b   description 4 desc 3 val         1    desc 4 val
# 10  b   description 5 desc 4 val         1          <NA>

Using the shift_val which is associated with each unique df$ID, I am able to shift the var_value up, where the value matches the var_description.

The current solution I am using works but when applied to a much larger dataframe (such as one that is ~ >100k rows), it becomes slow and inefficient.

Can someone recommend an alternative solution which gives me the same output as the one I have but has potential to be much more efficient? perhaps a data.table or purrr solution might be best, but I am not well versed in either. Any help is appreciated!