Replacing NA's with a unique sequence in a list of dataframes-CodePudding

I have example data as follows:

library(data.table)
datA <- fread("ID somevar
               NA 4
               NA  3
               2  5")
datB <- fread("ID somevar
               7   4
               NA  3
               NA  5")
dat_list <- list(datA, datB)

In dat_list I would like to replace all NA's in the ID column with a new ID-number.

I would like this number to be unique and start at 100. I thought of something like this:

for (i in seq_along(dat_list)){
  temp <- dat_list[[i]]
  count_of_seq <- sum(is.na(temp$ID))
  sequence_dat <- seq(100, 100 count_of_seq)
  temp <- setDT(temp)[is.na(ID), ID:=sequence_dat[i]]
}

But this does not work because, it uses only one number of the sequence for each list:

How should I do this properly?

Desired output:

library(data.table)
datA <- fread("ID somevar
               100 4
               101  3
               2  5")
datB <- fread("ID somevar
               7   4
               102  3
               103  5")
dat_list <- list(datA, datB)

CodePudding user response：

One dplyr option could be:

dat_list %>%
 bind_rows(., .id = "dataset_ID") %>%
 mutate(ID = ifelse(is.na(ID), 99   cumsum(is.na(ID)), ID)) %>%
 group_split(dataset_ID, .keep = FALSE)

[[1]]
# A tibble: 3 × 2
     ID somevar
  <dbl>   <int>
1   100       4
2   101       3
3     2       5

[[2]]
# A tibble: 3 × 2
     ID somevar
  <dbl>   <int>
1     7       4
2   102       3
3   103       5

CodePudding user response：

In base R:

id <- 99
for (i in seq_along(dat_list)) {
  nas <- is.na(dat_list[[i]]$ID)
  dat_list[[i]]$ID[nas] <- id   seq_len(sum(nas))
  id <- id   sum(nas)
}

dat_list
# [[1]]
#     ID somevar
# 1: 100       4
# 2: 101       3
# 3:   2       5
# 
# [[2]]
#     ID somevar
# 1:   7       4
# 2: 102       3
# 3: 103       5