Chapter 34 Writing Data
What You’ll Learn:
- Writing CSV and text files
- Excel export
- Common export errors
- File permissions
- Data serialization
- Format preservation
Key Errors Covered: 18+ export errors
Difficulty: ⭐⭐ Intermediate
34.2 Writing CSV Files
💡 Key Insight: Base R vs readr
# Create sample data
data <- mtcars[1:5, 1:5]
# Base R
write.csv(data, "base_output.csv")
# - Adds row names by default
# - Quotes strings
# - Slower
# readr
write_csv(data, "readr_output.csv")
# - No row names
# - Only quotes when needed
# - Faster
# - Better handling of special values
# Compare
cat("Base R file:\n")
#> Base R file:
cat(paste(head(readLines("base_output.csv"), 3), collapse = "\n"))
#> "","mpg","cyl","disp","hp","drat"
#> "Mazda RX4",21,6,160,110,3.9
#> "Mazda RX4 Wag",21,6,160,110,3.9
cat("\n\nreadr file:\n")
#>
#>
#> readr file:
cat(paste(head(readLines("readr_output.csv"), 3), collapse = "\n"))
#> mpg,cyl,disp,hp,drat
#> 21,6,160,110,3.9
#> 21,6,160,110,3.934.3 Error #1: cannot open the connection
⭐ BEGINNER 📁 PATH
34.3.1 The Error
# Try to write to non-existent directory
write_csv(mtcars, "nonexistent_dir/file.csv")
#> Error: Cannot open file for writing:
#> * 'nonexistent_dir/file.csv'🔴 ERROR
Error: Cannot open file for writing:
'nonexistent_dir/file.csv'34.3.3 Solutions
✅ SOLUTION 1: Create Directory First
# Check if directory exists
output_dir <- "output"
if (!dir.exists(output_dir)) {
dir.create(output_dir, recursive = TRUE)
cat("Created directory:", output_dir, "\n")
}
#> Created directory: output
# Now write
write_csv(mtcars, file.path(output_dir, "data.csv"))
cat("File written successfully\n")
#> File written successfully✅ SOLUTION 2: Safe Write Function
safe_write_csv <- function(data, path, ...) {
# Get directory
dir_path <- dirname(path)
# Create if doesn't exist
if (!dir.exists(dir_path)) {
dir.create(dir_path, recursive = TRUE)
message("Created directory: ", dir_path)
}
# Check write permission
if (!file.access(dir_path, mode = 2) == 0) {
stop("No write permission for directory: ", dir_path)
}
# Write file
write_csv(data, path, ...)
# Verify
if (file.exists(path)) {
message("Successfully wrote ", nrow(data), " rows to ", path)
} else {
stop("File was not created")
}
invisible(path)
}
# Test
safe_write_csv(mtcars, "test_output/cars.csv")
#> Created directory: test_output
#> Successfully wrote 32 rows to test_output/cars.csv34.4 Writing Options
💡 Key Insight: Write Options
# Control output format
write_csv(mtcars, "formatted.csv",
na = "MISSING", # How to write NAs
quote = "all") # Quote all fields
# Append to existing file
write_csv(mtcars[1:5, ], "append_test.csv")
write_csv(mtcars[6:10, ], "append_test.csv", append = TRUE)
cat("Lines in file:", length(readLines("append_test.csv")), "\n")
#> Lines in file: 11
# Write with semicolon delimiter (European)
write_csv2(mtcars, "european.csv")
# Custom delimiter
write_delim(mtcars, "pipe_delim.txt", delim = "|")
# Tab-separated
write_tsv(mtcars, "tab_separated.txt")34.5 Writing Excel Files
💡 Key Insight: writexl Package
library(writexl)
# Single sheet
write_xlsx(mtcars, "cars.xlsx")
# Multiple sheets
write_xlsx(
list(
Cars = mtcars,
Iris = iris,
Summary = data.frame(
Dataset = c("mtcars", "iris"),
Rows = c(nrow(mtcars), nrow(iris))
)
),
"multi_sheet.xlsx"
)
# Verify
library(readxl)
excel_sheets("multi_sheet.xlsx")
#> [1] "Cars" "Iris" "Summary"34.6 Error #2: File Permission Denied
⭐⭐ INTERMEDIATE 💻 SYSTEM
34.6.2 Solutions
✅ SOLUTION 1: Check File Access
check_file_writable <- function(path) {
# Check if file exists and is writable
if (file.exists(path)) {
if (file.access(path, mode = 2) == 0) {
cat("File is writable\n")
return(TRUE)
} else {
warning("File exists but is not writable (may be open)")
return(FALSE)
}
} else {
# Check if directory is writable
dir_path <- dirname(path)
if (file.access(dir_path, mode = 2) == 0) {
cat("Directory is writable\n")
return(TRUE)
} else {
warning("No write permission for directory")
return(FALSE)
}
}
}
# Test
check_file_writable("test.csv")
#> Directory is writable
#> [1] TRUE✅ SOLUTION 2: Use Temporary File
# Write to temp file first
temp_file <- tempfile(fileext = ".csv")
write_csv(mtcars, temp_file)
# Then copy/move to destination
final_path <- "output/final.csv"
if (!dir.exists("output")) dir.create("output")
file.copy(temp_file, final_path, overwrite = TRUE)
#> [1] TRUE
cat("File written via temporary location\n")
#> File written via temporary location34.7 Preserving Data Types
🎯 Best Practice: Preserve Types
library(lubridate)
# Create data with various types
complex_data <- tibble(
id = 1:3,
name = c("Alice", "Bob", "Charlie"),
value = c(1.5, 2.3, 3.7),
date = ymd("2024-01-01") + days(0:2),
logical = c(TRUE, FALSE, TRUE),
factor_col = factor(c("A", "B", "C"))
)
# CSV loses some type info
write_csv(complex_data, "types_csv.csv")
from_csv <- read_csv("types_csv.csv", show_col_types = FALSE)
str(from_csv)
#> spc_tbl_ [3 × 6] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
#> $ id : num [1:3] 1 2 3
#> $ name : chr [1:3] "Alice" "Bob" "Charlie"
#> $ value : num [1:3] 1.5 2.3 3.7
#> $ date : Date[1:3], format: "2024-01-01" "2024-01-02" ...
#> $ logical : logi [1:3] TRUE FALSE TRUE
#> $ factor_col: chr [1:3] "A" "B" "C"
#> - attr(*, "spec")=
#> .. cols(
#> .. id = col_double(),
#> .. name = col_character(),
#> .. value = col_double(),
#> .. date = col_date(format = ""),
#> .. logical = col_logical(),
#> .. factor_col = col_character()
#> .. )
#> - attr(*, "problems")=<externalptr>
# RDS preserves everything
saveRDS(complex_data, "types_rds.rds")
from_rds <- readRDS("types_rds.rds")
str(from_rds)
#> tibble [3 × 6] (S3: tbl_df/tbl/data.frame)
#> $ id : int [1:3] 1 2 3
#> $ name : chr [1:3] "Alice" "Bob" "Charlie"
#> $ value : num [1:3] 1.5 2.3 3.7
#> $ date : Date[1:3], format: "2024-01-01" "2024-01-02" ...
#> $ logical : logi [1:3] TRUE FALSE TRUE
#> $ factor_col: Factor w/ 3 levels "A","B","C": 1 2 3
# Excel preserves some types
write_xlsx(complex_data, "types_excel.xlsx")
from_excel <- read_excel("types_excel.xlsx")
str(from_excel)
#> tibble [3 × 6] (S3: tbl_df/tbl/data.frame)
#> $ id : num [1:3] 1 2 3
#> $ name : chr [1:3] "Alice" "Bob" "Charlie"
#> $ value : num [1:3] 1.5 2.3 3.7
#> $ date : POSIXct[1:3], format: "2024-01-01" "2024-01-02" ...
#> $ logical : logi [1:3] TRUE FALSE TRUE
#> $ factor_col: chr [1:3] "A" "B" "C"34.8 Large Data Export
🎯 Best Practice: Writing Large Files
# For very large data
library(data.table)
fwrite(large_data, "large_file.csv")
# Write in chunks
write_csv_chunked <- function(data, path, chunk_size = 10000) {
n_chunks <- ceiling(nrow(data) / chunk_size)
for (i in 1:n_chunks) {
start <- (i - 1) * chunk_size + 1
end <- min(i * chunk_size, nrow(data))
chunk <- data[start:end, ]
write_csv(chunk, path,
append = i > 1, # Append after first chunk
col_names = i == 1) # Only write headers first time
cat("Wrote chunk", i, "of", n_chunks, "\n")
}
}
# Compressed output
library(readr)
write_csv(data, gzfile("data.csv.gz"))34.9 R-Specific Formats
💡 Key Insight: R Binary Formats
# RDS - single object
saveRDS(mtcars, "mtcars.rds")
cars_rds <- readRDS("mtcars.rds")
# RData/rda - multiple objects
x <- 1:10
y <- "text"
save(x, y, mtcars, file = "data.RData")
rm(x, y, mtcars)
load("data.RData") # Restores objects with original names
# Feather - fast, interoperable
library(arrow)
#> Error in library(arrow): there is no package called 'arrow'
write_feather(mtcars, "mtcars.feather")
#> Error in write_feather(mtcars, "mtcars.feather"): could not find function "write_feather"
cars_feather <- read_feather("mtcars.feather")
#> Error in read_feather("mtcars.feather"): could not find function "read_feather"
# Parquet - compressed, columnar
write_parquet(mtcars, "mtcars.parquet")
#> Error in write_parquet(mtcars, "mtcars.parquet"): could not find function "write_parquet"
cars_parquet <- read_parquet("mtcars.parquet")
#> Error in read_parquet("mtcars.parquet"): could not find function "read_parquet"34.10 Error #3: Overwriting Files
⭐ BEGINNER ⚠️ SAFETY
34.10.2 Solutions
✅ SOLUTION: Safe Write with Backup
safe_write_with_backup <- function(data, path, backup = TRUE, ...) {
# If file exists and backup requested
if (file.exists(path) && backup) {
# Create backup
backup_path <- paste0(path, ".backup.",
format(Sys.time(), "%Y%m%d_%H%M%S"))
file.copy(path, backup_path)
message("Created backup: ", basename(backup_path))
}
# Write new file
write_csv(data, path, ...)
message("Wrote file: ", path)
invisible(path)
}
# Create test file
write_csv(mtcars[1:5, ], "important.csv")
# Safely overwrite
safe_write_with_backup(mtcars[1:10, ], "important.csv")
#> Created backup: important.csv.backup.20251026_154714
#> Wrote file: important.csv
# Check backups
list.files(pattern = "important")
#> [1] "important.csv"
#> [2] "important.csv.backup.20251026_154714"34.11 Format Comparison
🎯 Best Practice: Choose Right Format
# Compare formats
compare_formats <- function(data) {
formats <- list(
CSV = function() write_csv(data, "test.csv"),
RDS = function() saveRDS(data, "test.rds"),
Excel = function() write_xlsx(data, "test.xlsx"),
Feather = function() write_feather(data, "test.feather"),
Parquet = function() write_parquet(data, "test.parquet")
)
results <- data.frame(
Format = names(formats),
Size_KB = numeric(length(formats)),
Time_ms = numeric(length(formats))
)
for (i in seq_along(formats)) {
# Time it
time <- system.time(formats[[i]]())["elapsed"] * 1000
# Get size
files <- list.files(pattern = "^test\\.")
size <- sum(file.info(files)$size) / 1024
results$Size_KB[i] <- round(size, 2)
results$Time_ms[i] <- round(time, 2)
# Cleanup
unlink(files)
}
results
}
# Test with mtcars
compare_formats(mtcars)
#> Error in write_feather(data, "test.feather"): could not find function "write_feather"
#> Timing stopped at: 0.001 0 034.12 Summary
Key Takeaways:
- Create directories first - Check with dir.exists()
- Use readr functions - write_csv(), not write.csv()
- Check permissions - Verify file access
- Preserve types - Use RDS for R objects
- Backup important files - Before overwriting
- Choose right format - CSV for sharing, RDS for R
- Handle large files - Use data.table or chunks
Quick Reference:
| Format | Function | Best For |
|---|---|---|
| CSV | write_csv() | Sharing, Excel |
| RDS | saveRDS() | R objects, types |
| Excel | write_xlsx() | Multiple sheets |
| Feather | write_feather() | Fast, interoperable |
| Parquet | write_parquet() | Big data, compressed |
Writing Functions:
# CSV
write_csv(data, "file.csv")
write_csv2(data, "european.csv") # Semicolon
write_tsv(data, "file.txt") # Tab
write_delim(data, "file.txt", "|")
# Excel
library(writexl)
write_xlsx(data, "file.xlsx")
write_xlsx(list(Sheet1 = data1, Sheet2 = data2), "file.xlsx")
# R formats
saveRDS(data, "file.rds")
save(obj1, obj2, file = "file.RData")
# Modern formats
library(arrow)
write_feather(data, "file.feather")
write_parquet(data, "file.parquet")Best Practices:
# ✅ Good
Create directories before writing
Check file permissions
Use write_csv() not write.csv()
Backup before overwriting
Choose appropriate format
Compress large files
Test write with small sample first
# ❌ Avoid
Writing to non-existent directories
Overwriting without backup
Using write.csv() for new code
Writing very large files without chunks
Ignoring permission errors