How to compute the NAs with the column mean and then multiply columns of different lengths in R?

Question

My question might be not so clear so I am putting an example.

My final goal is to produce

final=(df1$a*df2$b)+(df1$a*df3$c*df4$d)+(df4$d*df5$e)

I have five data frames (one column each) with different lengths as follows:

df1

    a
1.  1
2.  2
3.  4
4.  2

df2

    b
1.  2
2.  6

df3

    c
1.  2
2.  4 
3.  3

df4

    d
1.  1
2.  2
3.  4
4.  3

df5

    e
1.  4
2.  6
3.  2

So I want a final database which includes them all as follows

finaldf

    a   b   c   d  e
1.  1   2   2   1  4
2.  2   6   4   2  6
3.  4   NA  3   4  2
4.  2   NA  NA  3  NA

I want all the NAs for each column to be replaced with the mean of that column, so the finaldf has equal length of all the columns:

finaldf

    a   b   c   d   e
1.  1   2   2   1   4
2.  2   6   4   2   6
3.  4   4   3   4   2
4.  2   4   3   3   4

and therefore I can produce a final result for final=(df1$a*df2$b)+(df1$a*df3$c*df4$d)+(df4$d*df5$e) as I need.

Answer 1

The easiest by far is to use the qpcR, dplyr and tidyr packages.

library(dplyr)
library(qpcR)
library(tidyr)

df1 <- data.frame(a=c(1,2,4,2))
df2 <- data.frame(b=c(2,6))
df3 <- data.frame(c=c(2,4,3))
df4 <- data.frame(d=c(1,2,4,3))
df5 <- data.frame(e=c(4,6,2))

mydf <- qpcR:::cbind.na(df1, df2, df3, df4,df5) %>% 
  tidyr::replace_na(.,as.list(colMeans(.,na.rm=T)))

> mydf
  a b c d e
1 1 2 2 1 4
2 2 6 4 2 6
3 4 4 3 4 2
4 2 4 3 3 4

Depending on your rgl, settings, you might need to run the following at the top of your script to make the qpcR package load (see https://stackoverflow.com/a/66127391/2554330 ):

options(rgl.useNULL = TRUE)
library(rgl)

Answer 2

With purrr and dplyr, we can first put all dataframes in a list with mget(). Second, use set_names to replace the dataframe names with their respective column names. As a third step, unlist the dataframes to get vectors with pluck . Then add the NAs by making all vectors the same length . Finally, bind all vectors back into a dataframe with as.data.frame , then use mutate with ~replace_na and colmeans.

mget(ls(pattern = 'df\\d')) %>%
        set_names(map_chr(., colnames)) %>%
        map(pluck, 1) %>%
        map(., `length<-`, max(lengths(.))) %>%
        as.data.frame %>%
        mutate(across(everything(), ~replace_na(.x, mean(.x, na.rm=TRUE))))