R Cran Read Csv Strings as Factor
Reading and Writing CSV Files
Overview
Teaching: 30 min
Exercises: 0 minQuestions
How practise I read data from a CSV file into R?
How do I write data to a CSV file?
Objectives
Read in a .csv, and explore the arguments of the csv reader.
Write the altered data fix to a new .csv, and explore the arguments.
The most mutual way that scientists store data is in Excel spreadsheets. While in that location are R packages designed to access data from Excel spreadsheets (e.1000., gdata, RODBC, XLConnect, xlsx, RExcel), users often observe it easier to salve their spreadsheets in comma-separated values files (CSV) and so use R's built in functionality to read and manipulate the information. In this short lesson, we'll acquire how to read data from a .csv and write to a new .csv, and explore the arguments that allow y'all read and write the data correctly for your needs.
Read a .csv and Explore the Arguments
Let's kickoff by opening a .csv file containing data on the speeds at which cars of different colors were clocked in 45 mph zones in the four-corners states (CarSpeeds.csv). Nosotros volition use the congenital in read.csv(...) function call, which reads the information in as a data frame, and assign the data frame to a variable (using <-) and so that it is stored in R'southward memory. Then we volition explore some of the basic arguments that can be supplied to the role. First, open the RStudio projection containing the scripts and data you were working on in episode 'Analyzing Patient Data'.
# Import the data and look at the offset half dozen rows carSpeeds <- read.csv ( file = 'data/motorcar-speeds.csv' ) head ( carSpeeds ) Color Speed State i Blue 32 NewMexico two Red 45 Arizona 3 Blueish 35 Colorado iv White 34 Arizona five Red 25 Arizona 6 Blueish 41 Arizona Changing Delimiters
The default delimiter of the
read.csv()function is a comma, only you can use other delimiters by supplying the 'sep' argument to the function (eastward.g., typingsep = ';'allows a semi-colon separated file to be correctly imported - see?read.csv()for more data on this and other options for working with different file types).
The phone call to a higher place will import the information, but we have not taken advantage of several handy arguments that can be helpful in loading the information in the format nosotros want. Let's explore some of these arguments.
The default for read.csv(...) is to set up the header argument to Truthful. This means that the first row of values in the .csv is set as header data (column names). If your data prepare does not accept a header, set the header argument to Faux:
# The get-go row of the data without setting the header argument: carSpeeds [ 1 , ] Colour Speed State 1 Bluish 32 NewMexico # The first row of the data if the header argument is set to Simulated: carSpeeds <- read.csv ( file = 'data/machine-speeds.csv' , header = Fake ) carSpeeds [ 1 , ] V1 V2 V3 one Color Speed State Conspicuously this is non the desired beliefs for this data set, but information technology may exist useful if you have a dataset without headers.
The stringsAsFactors Argument
In older versions of R (prior to four.0) this was perhaps the virtually important argument in read.csv(), particularly if you were working with categorical data. This is considering the default behavior of R was to convert character strings into factors, which may brand it hard to practice such things as supersede values. It is important to exist aware of this behaviour, which we will demonstrate. For example, permit'south say nosotros observe out that the data collector was color blind, and accidentally recorded green cars as being blue. In order to correct the data fix, let'south replace 'Blue' with 'Greenish' in the $Color column:
# Hither we will use R's `ifelse` function, in which we provide the exam phrase, # the outcome if the result of the exam is 'TRUE', and the upshot if the # result is 'FALSE'. We will too assign the results to the Colour cavalcade, # using '<-' # First - reload the data with a header carSpeeds <- read.csv ( file = 'data/automobile-speeds.csv' , stringsAsFactors = TRUE ) carSpeeds $ Color <- ifelse ( carSpeeds $ Color == 'Blue' , 'Dark-green' , carSpeeds $ Color ) carSpeeds $ Colour [ane] "Green" "1" "Dark-green" "5" "4" "Green" "Green" "2" "v" [10] "4" "4" "5" "Green" "Green" "two" "four" "Green" "Dark-green" [19] "v" "Green" "Light-green" "Light-green" "4" "Green" "4" "4" "4" [28] "four" "5" "Light-green" "4" "5" "2" "4" "2" "2" [37] "Green" "4" "2" "4" "two" "2" "4" "four" "5" [46] "2" "Green" "four" "4" "2" "ii" "4" "5" "four" [55] "Green" "Green" "2" "Green" "five" "2" "4" "Light-green" "Green" [64] "v" "2" "4" "four" "2" "Green" "5" "Green" "4" [73] "5" "v" "Green" "Green" "Greenish" "Greenish" "Green" "5" "2" [82] "Green" "5" "ii" "2" "iv" "4" "5" "5" "five" [91] "five" "four" "4" "iv" "v" "2" "5" "2" "2" [100] "5" What happened?!? It looks like 'Blue' was replaced with 'Green', just every other color was turned into a number (equally a character string, given the quote marks before and afterwards). This is because the colors of the cars were loaded as factors, and the factor level was reported following replacement.
To run into the internal structure, we can use another function, str(). In this instance, the dataframe's internal construction includes the format of each column, which is what nosotros are interested in. str() will be reviewed a little more in the lesson Data Types and Structures.
# Reload the data with a header (the previous ifelse call modifies attributes) carSpeeds <- read.csv ( file = 'data/automobile-speeds.csv' , stringsAsFactors = Truthful ) str ( carSpeeds ) 'data.frame': 100 obs. of three variables: $ Color: Factor westward/ five levels " Cherry","Blackness",..: 3 i 3 5 four iii 3 2 v 4 ... $ Speed: int 32 45 35 34 25 41 34 29 31 26 ... $ State: Cistron w/ iv levels "Arizona","Colorado",..: 3 1 two 1 1 i 3 two ane 2 ... Nosotros can run into that the $Colour and $Country columns are factors and $Speed is a numeric cavalcade.
At present, let'southward load the dataset using stringsAsFactors=Faux, and meet what happens when we try to supplant 'Blue' with 'Green' in the $Color column:
carSpeeds <- read.csv ( file = 'data/car-speeds.csv' , stringsAsFactors = Imitation ) str ( carSpeeds ) 'data.frame': 100 obs. of 3 variables: $ Color: chr "Bluish" " Red" "Blueish" "White" ... $ Speed: int 32 45 35 34 25 41 34 29 31 26 ... $ State: chr "NewMexico" "Arizona" "Colorado" "Arizona" ... carSpeeds $ Color <- ifelse ( carSpeeds $ Color == 'Blue' , 'Dark-green' , carSpeeds $ Colour ) carSpeeds $ Color [1] "Light-green" " Red" "Green" "White" "Red" "Greenish" "Green" "Blackness" "White" [10] "Red" "Crimson" "White" "Green" "Green" "Black" "Cherry-red" "Green" "Light-green" [nineteen] "White" "Green" "Greenish" "Light-green" "Red" "Dark-green" "Red" "Cherry-red" "Reddish" [28] "Cherry" "White" "Dark-green" "Red" "White" "Black" "Red" "Blackness" "Blackness" [37] "Green" "Red" "Black" "Ruby" "Black" "Black" "Red" "Red" "White" [46] "Black" "Green" "Cerise" "Crimson" "Black" "Black" "Red" "White" "Red" [55] "Green" "Green" "Blackness" "Green" "White" "Black" "Ruby" "Green" "Green" [64] "White" "Black" "Cherry" "Cerise" "Black" "Green" "White" "Dark-green" "Red" [73] "White" "White" "Green" "Green" "Green" "Green" "Green" "White" "Blackness" [82] "Green" "White" "Black" "Black" "Red" "Red" "White" "White" "White" [91] "White" "Red" "Red" "Ruby-red" "White" "Blackness" "White" "Black" "Black" [100] "White" That'south ameliorate! And we tin see how the data now is read as character instead of factor. From R version four.0 onwards nosotros practice not have to specify stringsAsFactors=FALSE, this is the default beliefs.
The as.is Argument
This is an extension of the stringsAsFactors statement, but gives you control over individual columns. For example, if nosotros want the colors of cars imported as strings, but nosotros want the names of the states imported as factors, we would load the information ready every bit:
carSpeeds <- read.csv ( file = 'information/car-speeds.csv' , every bit.is = 1 ) # Notation, the 1 applies equally.is to the showtime column only Now nosotros can meet that if nosotros try to replace 'Blueish' with 'Light-green' in the $Color column everything looks fine, while trying to replace 'Arizona' with 'Ohio' in the $State column returns the factor numbers for the names of states that we haven't replaced:
'data.frame': 100 obs. of iii variables: $ Colour: chr "Blue" " Cherry" "Blue" "White" ... $ Speed: int 32 45 35 34 25 41 34 29 31 26 ... $ State: Cistron westward/ four levels "Arizona","Colorado",..: 3 1 2 1 1 one 3 two 1 2 ... carSpeeds $ Color <- ifelse ( carSpeeds $ Color == 'Blue' , 'Green' , carSpeeds $ Color ) carSpeeds $ Color [1] "Light-green" " Blood-red" "Green" "White" "Red" "Green" "Light-green" "Black" "White" [ten] "Red" "Cherry" "White" "Green" "Dark-green" "Blackness" "Red" "Green" "Green" [19] "White" "Green" "Light-green" "Green" "Carmine" "Green" "Ruby-red" "Red" "Reddish" [28] "Red" "White" "Light-green" "Blood-red" "White" "Black" "Cherry" "Black" "Black" [37] "Light-green" "Scarlet" "Black" "Red" "Black" "Black" "Carmine" "Red" "White" [46] "Black" "Green" "Reddish" "Blood-red" "Blackness" "Black" "Red" "White" "Red" [55] "Dark-green" "Green" "Black" "Light-green" "White" "Black" "Ruby" "Green" "Green" [64] "White" "Black" "Red" "Blood-red" "Black" "Green" "White" "Green" "Red" [73] "White" "White" "Green" "Dark-green" "Dark-green" "Green" "Greenish" "White" "Blackness" [82] "Light-green" "White" "Black" "Black" "Red" "Red" "White" "White" "White" [91] "White" "Red" "Cherry" "Red" "White" "Black" "White" "Black" "Black" [100] "White" carSpeeds $ State <- ifelse ( carSpeeds $ State == 'Arizona' , 'Ohio' , carSpeeds $ State ) carSpeeds $ Land [1] "3" "Ohio" "ii" "Ohio" "Ohio" "Ohio" "3" "ii" "Ohio" "2" [eleven] "4" "4" "4" "4" "4" "3" "Ohio" "3" "Ohio" "iv" [21] "four" "four" "3" "2" "2" "3" "2" "four" "ii" "4" [31] "3" "2" "2" "iv" "two" "two" "three" "Ohio" "4" "ii" [41] "ii" "3" "Ohio" "four" "Ohio" "ii" "3" "three" "3" "2" [51] "Ohio" "iv" "4" "Ohio" "3" "2" "4" "two" "4" "4" [61] "4" "2" "3" "2" "3" "2" "3" "Ohio" "3" "4" [71] "iv" "2" "Ohio" "iv" "2" "ii" "2" "Ohio" "3" "Ohio" [81] "4" "ii" "2" "Ohio" "Ohio" "Ohio" "four" "Ohio" "four" "4" [91] "4" "Ohio" "Ohio" "3" "2" "2" "4" "3" "Ohio" "4" Nosotros tin see that $Color column is a character while $Country is a cistron.
Updating Values in a Factor
Suppose we want to keep the colors of cars as factors for some other operations we want to perform. Write lawmaking for replacing 'Blue' with 'Greenish' in the
$Colorcolumn of the cars dataset without importing the information withstringsAsFactors=FALSE.Solution
carSpeeds <- read.csv ( file = 'data/car-speeds.csv' ) # Replace 'Blue' with 'Greenish' in cars$Color without using the stringsAsFactors # or as.is arguments carSpeeds $ Colour <- ifelse ( as.character ( carSpeeds $ Color ) == 'Blue' , 'Green' , as.grapheme ( carSpeeds $ Colour )) # Convert colors back to factors carSpeeds $ Color <- as.cistron ( carSpeeds $ Color )
The strip.white Argument
It is not uncommon for mistakes to have been made when the information were recorded, for example a space (whitespace) may accept been inserted earlier a information value. By default this whitespace will exist kept in the R surround, such that '\ Red' volition be recognized as a dissimilar value than 'Red'. In lodge to avoid this blazon of fault, use the strip.white argument. Permit's come across how this works by checking for the unique values in the $Color column of our dataset:
Hither, the data recorder added a space earlier the colour of the automobile in one of the cells:
# We use the built-in unique() function to extract the unique colors in our dataset unique ( carSpeeds $ Color ) [i] Green Cherry-red White Red Black Levels: Red Blackness Green Red White Oops, we see two values for red cars.
Permit's try again, this time importing the data using the strip.white argument. Notation - this argument must be accompanied by the sep argument, by which we indicate the type of delimiter in the file (the comma for most .csv files)
carSpeeds <- read.csv ( file = 'data/car-speeds.csv' , stringsAsFactors = FALSE , strip.white = TRUE , sep = ',' ) unique ( carSpeeds $ Color ) [1] "Blue" "Ruby-red" "White" "Black" That's better!
Specify Missing Information When Loading
It is common for data sets to take missing values, or mistakes. The convention for recording missing values oftentimes depends on the individual who collected the data and tin exist recorded as
north.a.,--, or empty cells " ". R recognises the reserved character cordNAas a missing value, but not some of the examples to a higher place. Let's say the inflamation scale in the data set up we used earlierinflammation-01.csvactually starts at1for no inflamation and the zippo values (0) were a missed observation. Looking at the?read.csvhelp page is at that place an argument we could utilise to ensure all zeros (0) are read in every bitNA? Perhaps, in thecar-speeds.csvinformation contains mistakes and the person measuring the car speeds could not accurately distinguish between "Black or "Blue" cars. Is there a way to specify more than 1 'string', such every bit "Black" and "Bluish", to exist replaced pastNASolution
read.csv ( file = "data/inflammation-01.csv" , na.strings = "0" )or , in
automobile-speeds.csvuse a grapheme vector for multiple values.read.csv ( file = 'information/car-speeds.csv' , na.strings = c ( "Black" , "Blue" ) )
Write a New .csv and Explore the Arguments
After altering our cars dataset by replacing 'Blue' with 'Light-green' in the $Color column, nosotros at present want to salvage the output. There are several arguments for the write.csv(...) function phone call, a few of which are particularly important for how the data are exported. Let's explore these now.
# Export the data. The write.csv() role requires a minimum of 2 # arguments, the information to be saved and the name of the output file. write.csv ( carSpeeds , file = 'data/car-speeds-cleaned.csv' ) If you open up the file, you'll see that it has header names, considering the information had headers within R, but that at that place are numbers in the commencement column.
The row.names Argument
This argument allows us to set the names of the rows in the output data file. R's default for this argument is TRUE, and since it does non know what else to name the rows for the cars information set, it resorts to using row numbers. To correct this, we can set row.names to Faux:
write.csv ( carSpeeds , file = 'data/motorcar-speeds-cleaned.csv' , row.names = Simulated ) Now we see:
Setting Column Names
In that location is also a
col.namesargument, which tin can be used to set the column names for a data set without headers. If the data ready already has headers (e.grand., we used theheaders = Truthfulstatement when importing the data) and then acol.namesargument volition be ignored.
The na Argument
In that location are times when we desire to specify certain values for NAs in the data set (e.g., nosotros are going to pass the data to a program that only accepts -9999 as a nodata value). In this case, nosotros want to set the NA value of our output file to the desired value, using the na argument. Let's see how this works:
# Commencement, replace the speed in the third row with NA, by using an index (square # brackets to betoken the position of the value we desire to supplant) carSpeeds $ Speed [ three ] <- NA caput ( carSpeeds ) Color Speed Land one Bluish 32 NewMexico ii Red 45 Arizona 3 Blue NA Colorado iv White 34 Arizona 5 Red 25 Arizona 6 Blueish 41 Arizona write.csv ( carSpeeds , file = 'data/car-speeds-cleaned.csv' , row.names = Imitation ) Now nosotros'll set NA to -9999 when we write the new .csv file:
# Note - the na statement requires a string input write.csv ( carSpeeds , file = 'data/car-speeds-cleaned.csv' , row.names = FALSE , na = '-9999' ) And we see:
Key Points
Import data from a .csv file using the
read.csv(...)function.Understand some of the key arguments bachelor for importing the information properly, including
header,stringsAsFactors,equally.is, andstrip.white.Write information to a new .csv file using the
write.csv(...)officeUnderstand some of the key arguments available for exporting the information properly, such every bit
row.names,col.names, andna.
Source: https://swcarpentry.github.io/r-novice-inflammation/11-supp-read-write-csv/
0 Response to "R Cran Read Csv Strings as Factor"
Publicar un comentario