R: replacing NAs in a data.frame with values in the same position in another dataframe

Question

I have a dataframe with some NA values:

dfa <- data.frame(a=c(1,NA,3,4,5,NA),b=c(1,5,NA,NA,8,9),c=c(7,NA,NA,NA,2,NA))
dfa

I would like t

Answer 1

In the tidyverse, you can use purrr::map2_df, which is a strictly bivariate version of mapply that simplifies to a data.frame, and dplyr::coalesce, which replaces NA values in its first argument with the corresponding ones in the second.

library(tidyverse)

dfrepair %>% 
    mutate_all(as.numeric) %>%    # coalesce is strict about types
    map2_df(dfa, ., coalesce)

## # A tibble: 6 × 3
##       a     b     c
##   <dbl> <dbl> <dbl>
## 1     1     1     7
## 2     3     5     7
## 3     3     4     6
## 4     4     3     5
## 5     5     8     2
## 6     7     9     3

Answer 2

We can use Map from base R to do a columnwise comparison between the two datasets

dfa[] <- Map(function(x,y) {x[is.na(x)] <- y[is.na(x)]; x}, dfa, dfrepair)
dfa
#  a b c
#1 1 1 7
#2 3 5 7
#3 3 4 6
#4 4 3 5
#5 5 8 2
#6 7 9 3

Answer 3

dfa <- data.frame(a=c(1,NA,3,4,5,NA),b=c(1,5,NA,NA,8,9),c=c(7,NA,NA,NA,2,NA))
dfa
dfrepair <- data.frame(a=c(2:7),b=c(6:1),c=c(8:3))
dfrepair 
library(dplyr)
coalesce(as.numeric(dfa), as.numeric(dfrepair))

  a b c
1 1 1 7
2 3 5 7
3 3 4 6
4 4 3 5
5 5 8 2
6 7 9 3

As the code in dplyr is written in C++ it is faster in most cases. An other important advantage is that coalesce as well as many other dplyr functions are the same in SQL. Using dplyr you learn SQL by coding in R. ;-)

Answer 4

You can do:

dfa <- data.frame(a=c(1,NA,3,4,5,NA),b=c(1,5,NA,NA,8,9),c=c(7,NA,NA,NA,2,NA))
dfrepair <- data.frame(a=c(2:7),b=c(6:1),c=c(8:3))
dfa[is.na(dfa)] <- dfrepair[is.na(dfa)]
dfa

  a b c
1 1 1 7
2 3 5 7
3 3 4 6
4 4 3 5
5 5 8 2
6 7 9 3