read.csv / write.csv examples for KC R Users group, 2014-01-18

R-IO-readcsv-writecsv.R

# read.csv / write.csv examples described here:
# http://earlglynn.github.io/R/input-output/readcsv-writecsv/index.html

#setwd("")   # Set working directory if necessary

# Examples
# 1. Read Sample.csv input file with read.csv.
# 2. Explain some details of data.frame
# 3. Write Sample-Copy.csv output file with write.csv.
# 4. Compare Sample.csv to Sample-Copy.csv.

# efg, 2014-01-15

############################################################
### Read csv file into data.frame

rawdata <- read.csv("Sample.csv")
rawdata

dim(rawdata)   # dimensions
nrow(rawdata)  # number of rows
ncol(rawdata)  # number of columns

names(rawdata)      # column names
row.names(rawdata)

str(rawdata)   # structure

############################################################
### Addressing elements of data.frame

# Column of data
rawdata$Grade
rawdata$Age
rawdata$Score   # typo here!

# Indexing like matrix:  [rows, columns]
rawdata[2,4]    # single element: row 2, column 4
rawdata[2,]     # row 2
rawdata[,4]     # column 4

rawdata[1:2,c(1,3,5)]  # first 2 rows of columns 1, 3 and 5

### Lookup:  associate memory, hash, dictionary

# Single element
rawdata["2", "Amount"]   # row name "2", "Amount" column

# Rows with names "2" and "1", columns with names
# "Start", "Code" and "Age"
rawdata[c("2","1"), c("Start", "Code", "Age")]

############################################################
### Suppress factors:  stringsAsFactors=FALSE

# "Factors" are stored as integers internally
rawdata$Code
as.integer(rawdata$Code)
as.character(rawdata$Code)

# Re-read using stringsAsFactors=FALSE
rawdata <- read.csv("Sample.csv", stringsAsFactors=FALSE)
rawdata
str(rawdata)

############################################################
### Control data types by column:  colClasses

rawdata <- read.csv("Sample.csv",
           colClasses=c("factor", "integer", "character",
                        "numeric", "Date"))
str(rawdata)

# Data type suggestions:
# * Use "factor" for small, well-defined list of discrete values.
# * Use "character" for strings that may need to be
#     edited/modified/searched.
# * In vectors integers take 4 bytes, numeric doubles take 8 bytes.
#   E.g., for N <- 1000000L,
#     object.size(rep(0L,N))  is 4000000 + overhead,
#     object.size(rep(0.0,N)) is 8000000 + overhead,
#     where overhead is 24 bytes in 32-bit R and 40 bytes in 64-bit R.
# * The "best" format for Dates is ISO 8601.

############################################################
### Selected rows and columns

selected <- read.csv("Sample.csv", nrow=2, header=TRUE,
                     colClasses=c("NULL","integer","NULL",
                                  "numeric","NULL"))
selected

############################################################
### write.csv

# Default write.csv
write.csv(rawdata)  # echo to screen to review
write.csv(rawdata, "Sample-Copy-Default.csv")

# Drop row names.
write.csv(rawdata, row.names=FALSE)  # echo to screen
write.csv(rawdata, "Sample-Copy-No-RowNames.csv", row.names=FALSE)

# Exactly the same as original file (for this case)
write.csv(rawdata, row.names=FALSE, quote=FALSE, na="") # echo to screen
write.csv(rawdata, "Sample-Copy-Exact.csv", row.names=FALSE, quote=FALSE, na="")