• A stumbling block for many learning R is the Error Messages
• We often see them while we're loading data
• R is very strict about data formats
• We can load .xlsx, xls, csv, txt, gtf/gff files + many more
• The structure of the spreadsheet is vital

• Things we do to make it "look nice" in Excel can create problems
• In data, open RealTimeData.xlsx in Excel (or Libre Office)

Which sheet do you think will be the most problematic to load?

Sheet 1

• This is actually the type of format R loves to see
  • Simple column structure, with column and row names
  • No blank rows at the top or separating sub-tables
  • No blank columns
  • No merged cells

Sheet 2

What about all those missing values?

Sheet 2

What about all those missing values?

• R can happily deal with missing values: \(\implies\) will load as NA
• The missing column names may give strange behaviours
• Otherwise no problems

Sheet 3

• Here we effectively have 2 tables on the same sheet
  • Will this cause error messages?
  • R guesses the number of columns from the first row
  • The plot will simply be ignored

Always think in terms of columns

• In today's data folder is the file toothData.csv
• Clicking on it will open it in the Script Window as a text file
• We can use the button as shown

1. Try changing a few settings & note changes in the Preview
2. Once you're happy, click "Import"

You will see two lines of code in the Console \(\implies\) two things have just happened

toothData <- read.csv("data/toothData.csv")
View(toothData)

toothData <- read.csv("data/toothData.csv")

ALWAYS copy the first line into your script!

• This is the exact command we've used to load the file
• Anything we've set by clicking will also be in this line

View(toothData)

The second line has opened a preview of our R object

• By default, the R object will be named using the file-name before the .csv
• Also look at the toothData object in the Environment tab (click the arrow)

• The object toothData is known as a data.frame
• R equivalent to a spreadsheet
• Three possible ways to inspect this are:

View(toothData)
toothData
head(toothData)

What were the differences between each method?

• In a data.frame each column is a vector:
  • Each column has a single type of data
  • Different columns can be different types of data

• By default R assumes that a column of text is a categorical variable (i.e. a factor)
• We can change this by un-checking the stringsAsFactors button during import

• We can use the square braces to select rows and columns [row, column]

toothData[1:2,] # Select the first 2 rows, all columns
toothData[1:5, 1] # The first 5 entries in the first column

Because each column is a vector, we can use an alternative method ($)

toothData$len #Print the entire column vector called 'len'
toothData$len[1:5] # Now just the first 5 entries

Calling by name:

toothData[1:5, "len"]
toothData$len[1:5] 

When would you use either method?

One easy trap to fall into - forgetting the comma

toothData["len"]
toothData[["len"]]

• In the above section we called the R function read.csv()
• This is in the utils package which is one of the default packages
• A package is just a collection of related functions, e.g.
  • median() & sd() are in the stats package
  • max(), sum(), + are in the base package

• The package readr has a similar, but slightly superior version called read_csv()
• No GUI for this method

How do we install packages?

• There is a central repository known as CRAN

Tools > Install Packages ...

• Enter the package name readr then Install

• It should auto-complete once you start typing

• Now try the following packages: readxl, dplyr, ggplot2, reshape2, stringr

(This may take a few minutes - use the coloured post-it notes)

library(dplyr)
library(readr)

You might see lots of friendly messages…

toothData <- read_csv("data/toothData.csv")

1. Much faster
2. Character columns are left as plain text
3. It also runs some checks on your data \(\implies\) can give more error messages
4. Output is a local data frame (i.e. a tibble)
   \(\implies\) display in the Console is more convenient

toothData

?read_csv

• Once again we have four functions but stick to read_csv()
• This function has numerous arguments given as names
  (e.g. file, col_names)
  • If just the name is given, we need to specify something
  • Otherwise the argument is given a default value
    (e.g. col_names = TRUE)
  • If specifying in order, we don't need to name arguments

toothData <- read_csv("data/toothData.csv")

Is equivalent to:

toothData <- read_csv(file = "data/toothData.csv")

If we had a file with 3 blank lines to start, is there an argument that can help?

If we had a file with 3 blank lines to start, is there an argument that can help?

read_csv("path/to/file.csv", skip = 3)

• The bottom three functions are simplified wrappers to read_delim()
• read_csv() calls read_delim() using delim = ","
• read_csv2() calls read_delim() using delim = ";"
• read_tsv() calls read_delim() using delim = "\t"

What function would we call for space-delimited files?

R also has a package for loading .xls and xlsx files.

library(readxl)

The main function is read_excel()

?read_excel

• Missing values might be given a value (e.g. 9999, "NA")
• Column names might be missing
• File may have comments
• May be structural errors in the file
• White-space in cells
• Merged Cells (these are difficult…)

• read_csv() has an argument called col_names
• NB:read.csv() refers to column names as a header
• By default, the first row is assumed to contain the names of the variables/columns, i.e. col_names = TRUE
• This tells R how many columns you have

What happens if we get this wrong?

no_header <- read_csv("data/no_header.csv")

We can easily fix this

no_header <- read_csv("data/no_header.csv", col_names = FALSE)

What about that first column?

We can specify what is loaded or skipped using col_types

?read_csv

• We can skip columns using the - symbol
• character and numbers use c and n respectively

no_header <- read_csv("data/no_header.csv", col_names = FALSE,
                      col_types = "-ccnnc")

What if we get that wrong?

Let's mis-specify the third column as a number

no_header <- read_csv("data/no_header.csv", col_names = FALSE,
                      col_types = "-cnnnc")

• Did the error message make any sense?
• Did the file load?
• What happened to the third column?

Let's get it wrong first

comments <- read_csv("data/comments.csv")

Now we can get it right

comments <- read_csv("data/comments.csv", comment = "#")

This will work if there are comments in any rows

What happens when you try to load the file bad_colnames_.csv

bad_colnames <- read_csv("data/bad_colnames.csv")

How could we fix this?

1. By editing the file, and
2. Without editing the file

Here's my fix

bad_colnames <- read_csv("data/bad_colnames.csv", 
                             skip =  1, col_names = FALSE)
colnames(bad_colnames) <- c("rowname", "gender", "name",
                                "weight", "height", "transport")

• Here we have set the column names manually using a character vector
• Note that all words were quoted.

The most common function in R is c()

• This stands for combine
• Combines all values into a single R object, or vector
• If left empty, it is equivalent to NULL

c()

## NULL

colnames(bad_colnames) <- c()

What if missing values have been set to "-"?

Let's get it wrong first

missing_data <- read_csv("data/missing_data.csv")

Where have the errors appeared?

Now we can get it right

missing_data <- read_csv("data/missing_data.csv", na = "-")

After we've edited a file, we might wish to export it

?write_csv

• This is a wrapper for write_delim()
• Can export .csv, .txt, .tsv etc
• Individual R objects can be exported using write_rds()