0

I have made a function that read input from file lists and appends it using rbind.

dat <- NA
file.names <- list.files(paste(in.path2,"CSV",sep =""))
for(f in file.names){
  file <- paste(in.path2,"CSV/", f, sep = "")
  tmp <- read.csv(file, stringsAsFactors = F, na.strings = c("", " "))

  if (is.na(dat)) {
    dat <- tmp
  } else {
    colnames(tmp) <- colnames(dat)
    dat <- rbind(dat, tmp)
  }
  print(f)
}

I am getting this warning:

 1: In if (is.na(dat)) { ... :
  the condition has length > 1 and only the first element will be used.

How to correct this?

Shubh
  • 585
  • 9
  • 29
  • We an do this with `ifelse` instead of `if/else` Looks like you are checking for the NA in the whole dataset. it returns a logical matrix with `is.na` from the second run. Can you show a small reproducible example and what you expect – akrun Feb 14 '18 at 06:44

2 Answers2

1

I would strongly suggest to not grow your data.frame like that but like this

file.names <- list.files(paste(in.path2,"CSV",sep =""))
input_list <- list()
for(f in file.names){
  file <- paste(in.path2,"CSV/", f, sep = "")
  input_list[[f]] <- read.csv(file, stringsAsFactors = F, na.strings = c("", " "))
  print(f)
}
dat <- do.call(rbind, input_list)

This is much faster and you do not need to test if dat is.na or not

Linus
  • 705
  • 1
  • 10
  • 20
  • Each file has different column names .The rbind is returning error.There are 50 columns in each file having the same meaning but different colnames – Shubh Feb 14 '18 at 07:04
  • @Shubh `colnames(input_list[[f]]) <- 1:ncol(input_list[[f]])` in the for loop and at the end `colnames(dat) <- c(your, prefarred, names)` – Linus Feb 14 '18 at 07:09
1

We can do this with more easily with lapply without worrying about the assignment of NA and the if/else clauses

filenames <- list.files(paste0(in.path2,"CSV"), full.names = TRUE)
do.call(rbind,lapply(filenames, read.csv, na.strings = c("", " "), stringsAsFactors = FALSE))

Or another option is fread from data.table

library(data.table)
rbindlist(lapply(filenames, fread, na.strings = c("", " ")), fill = TRUE)

Or with tidyverse

library(tidyverse)
map_df(filenames, read_csv, na = c("", " "))

If the columns are not the same, then 

map(filenames, read_csv, na = c("", " ")) %>%
      bind_rows
akrun
  • 874,273
  • 37
  • 540
  • 662