如何在R中的空数据框中添加行

Question

I've tried everything but I can't add rows to an empty Data Frame. I realize that the first row its added but from the second to the end I get this error: invalid factor level, NA generated. I hope you can help me!! Thank you very much for your help!

   Table <- data.frame()

   for (i in 1:length(dfMoviesList)){

       ID = paste0("DF",i)
       Value = (dfMoviesList[[i]]$TITULO[1])
       Table <- rbind(Table,c(ID,Value)) 
   }

Answer 1

You could set stringsAsFactors=FALSE before running the code

op <- options(stringsAsFactors=FALSE)
Table <- data.frame()
for (i in 1:length(dfMoviesList)){
     ID = paste0("DF",i)
    Value = (dfMoviesList[[i]]$TITUL0[1])
    Table <- rbind(Table,c(ID,Value)) 
 }
options(op) #changes to default settings

Update

If speed is an issue, you could also try

ID <- paste0('DF', seq_along(dfMoviesList))
res <- data.frame(ID, Value=vapply(dfMoviesList, 
          function(x) x$TITUL0[1], numeric(1L))

data

set.seed(24)
 dfMoviesList <- lapply(1:3, function(i) 
    data.frame(TITUL0= sample(1:5), val=rnorm(5)) )

Benchmarks

set.seed(24)
dfMoviesList <- lapply(1:10000, function(i) 
        data.frame(TITUL0= sample(1:5)))

akrun1 <- function() { ID <- paste0('DF', seq_along(dfMoviesList))
                   data.frame(ID, Value=vapply(dfMoviesList, 
             function(x) x$TITUL0[1], numeric(1L)))
                    }
#included a data.table solution also
library(data.table)
akrun2 <- function() {DT <-  rbindlist(setNames(dfMoviesList, 
      paste0('DF', seq_along(dfMoviesList))), idcol=TRUE)
                     DT[DT[, .I[1L], .id]$V1]}   


dariober <- function(){
  Table<- matrix(nrow= length(dfMoviesList), ncol= 2, data= NA)

  for (i in 1:length(dfMoviesList)){
       ID<- paste0("DF",i)
       Value<- dfMoviesList[[i]]$TITUL0[1]
       Table[i,]<- c(ID, Value)
   }
 Table<- data.frame(ID= Table[,1], Value= Table[,2])

 }

 library(microbenchmark)

 microbenchmark(akrun1(), akrun2(), dariober(), times=20L, 
          unit='relative')
 #Unit: relative
 #     expr      min       lq     mean   median       uq      max neval cld
 #   akrun1() 2.214390 2.193538 2.055775 2.173440 2.148606 1.615028    20  b 
 #   akrun2() 1.000000 1.000000 1.000000 1.000000 1.000000 1.000000    20 a  
 # dariober() 3.226717 3.198742 2.984970 3.174609 3.139982 2.189399    20   c

Answer 2

Since you know before hand how many rows and columns you will have in your final data.frame, it's much faster to initialize an empty matrix of the right size, fill it, and convert it to data.frame.

In the for loop you propose the Table object is dropped and created at every iteration and this gets really slow even when the number of loops is not too large. See for example:

A sample of 10000 Movies:

dfMoviesList <- lapply(1:10000, function(i) 
    data.frame(TITUL0= sample(1:5)))

Empty matrix startegy:

system.time({
Table<- matrix(nrow= length(dfMoviesList), ncol= 2, data= NA)

for (i in 1:length(dfMoviesList)){
    ID<- paste0("DF",i)
    Value<- dfMoviesList[[i]]$TITUL0[1]
    Table[i,]<- c(ID, Value)
}
Table<- data.frame(ID= Table[,1], Value= Table[,2])
})
   user  system elapsed 
  0.129   0.001   0.130 

Compare to:

system.time({
op <- options(stringsAsFactors=FALSE)
Table <- data.frame()
for (i in 1:length(dfMoviesList)){
    ID = paste0("DF",i)
    Value = (dfMoviesList[[i]]$TITUL0[1])
    Table <- rbind(Table,c(ID,Value)) 
}
options(op)
})
   user  system elapsed 
 12.316   2.855  15.180