Transform a list of list (containing matrices) into a data.frame-CodePudding

I have a list of list which I want to transform into a data.frame. Some of the element in the "inner" list contains a matrix and I want to preserve columns and their name.

I made quite some progress using lapply, and now I ended up with another list which has this pattern:

[[1]]
[[1]]$X
[1] 0.100

[[1]]$Y
                   VAL   LCI  UCI  
AB                 1000  500  1300 
CD                 30    10   400

This pattern repeat for like 200 rows.

Now the question is that I want to create a data.frame with 4 columns: X, VAL, LCI, UCI. Obviously I would have rows which alternate "AB" and "CD" since these rownames are fixed in my list.

Anyone has an idea on how to perform this? Here is a reproducible example - although not fancy, list1 produced by this code is exactly the object that I need to transform in a dataframe:

#Create list x
var1 <- 0.100 
var1 <- list(var1)
names(var1) <- "x"

#Create list y
var2 <- matrix(1:6, nrow = 2)
rownames(var2) <- c("AB","CD")
colnames(var2) <- c("VAL","LCI","UCI")
var2 <- list(var2)
names(var2) <- "y"

#Create a second element
var3 <- 0.200
var3 <- list(var3)
names(var3) <- "x"

var4 <- matrix(7:12, nrow = 2)
rownames(var4) <- c("AB","CD")
colnames(var4) <- c("VAL","LCI","UCI")
var4 <- list(var4)
names(var4) <- "y"

#Create a list
list1 <- list(c(var1,var2),c(var3,var4))

CodePudding user response：

Not sure I completely understand your request but I give it a try! So, based on your example, please find below a reprex

Reprex

Code of the function DFbindRows

DFbindRows <- function(x){
  x <- do.call(rbind, lapply(x, as.data.frame))
  names(x) <- gsub("y.","",names(x))
  return(x)
}

Test of the function on your data list1

DF <- DFbindRows(list1)
#>       x   VAL   LCI   UCI
#> AB  0.1     1     3     5
#> CD  0.1     2     4     6
#> AB1 0.2     7     9    11
#> CD1 0.2     8    10    12

NB: the number after AB or CD is mandatory because it is not allowed to have duplicated row names in a dataframe

Check of the object class

class(DF)
#> [1] "data.frame"

^{Created on 2021-11-10 by the reprex package (v2.0.1)}

Workaround for the row names problem

To circumvent the problem I pointed out in the Nota Bene above, it is possible

to create an ID column in which the rows AB and CD alternate, and
to set numbers as row.names.

So, please find below the function DFbindRows2 which offers you the possibility to get such a dataframe

Reprex

Code of the function DFbindRows2

DFbindRows2 <- function(x){
  x <- do.call(rbind, lapply(x, as.data.frame))
  x$ID <- rep(c("AB", "CD"), nrow(x)/2)
  x <- x[,c("ID", 
            names(x)[-grep("ID",names(x))])]
  row.names(x) <- seq(nrow(x))
  names(x) <- gsub("y.","",names(x))
  return(x)
}

Test of the function on your data list1

DF <- DFbindRows2(list1)
#>   ID   x   VAL   LCI   UCI
#> 1 AB 0.1     1     3     5
#> 2 CD 0.1     2     4     6
#> 3 AB 0.2     7     9    11
#> 4 CD 0.2     8    10    12

Check of the object class

class(DF)
#> [1] "data.frame"

^{Created on 2021-11-10 by the reprex package (v2.0.1)}