In stenw/BST02R: Supporting Material for the R Course of the NIHES (BST02)

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Importing data files

If you want to learn to use R that is probably because you want to use it to analyse some data. So let's illustrate R based on a data set. We can download and import an SPSS data set containing data on an (imaginary) trial using:

library(foreign)
mydata <- read.spss(file='https://github.com/stenw/bst02r/blob/master/data-raw/pain.sav?raw=true',
                      to.data.frame = TRUE)

data.frames

When we have executed the command above the data set will be stored under the name mydata. mydata is a so called data.frame. This is a rectangular table in which every column contains a variable and every row an observation. The variable mydata will also appear in the environment tab in the upper right hand corner of RStudio. Right next to the name we can see how many observations and how many variables the data.frame has. We can examine the data set more closely by clicking the little icon next to it in this tab. An other way of looking at the data is my typing in the command View(mydata) directly in the Console tab.

We see that the first four variables are the patient number, the treatment, the gender and age of the patient(ptno, treatm, sex and age). In order to work with one of the variables in the data.frame we first type the name of the data.frame followed by a dollar sign, and finally the name of the variable inside the data.frame. So when we execute the command: mydata$age in the console the ages of all patients are displayed. We can calculate the average of them using the function mean.

mean(mydata$age)

vectors and datatypes

Each variable in the data.frame mydata is a so called vector. A vector is a number of values that each have the same data type (Or mode^[mode, storage.mode and type are closely related concepts, we will not discuss the differences here. See also ?mode.]). The most important data types in R are:^[There are a few more like complex and raw which we will not discuss.]:

------------ --------------------------------------------------- mode description ------------ ---------------------------------------------------
character: text, for example 'man', 'woman', 'censored', etc. logical: TRUE and FALSE integer: whole numers like 0, 1, 2, etc ... numeric: Posibly fractional numerical values like 1.0, 1.2 or 1e12 (that is 10 raised to the power of 12) -------------------------------------------------------------------

Vectors of these data types are the most elementary data structure in R. All other structures (like the data.table) are constructed using these vectors. In R there is also no structure that is smaller than a vector. A single number is not treated differently from a numeric vector of length ten; In fact R sees the single number simply as a numeric vector of length 1. The length() of a vector can be obtained by using the function length().

A vector can be created using the function c(). (The c stands for 'concatenate', 'coerce' or 'combine')

c(1, 2, 3)
c('spam', 'ham', 'eggs')
c("double", "quotes", "work",
  'like', 'single')
c(TRUE, FALSE)

In the output we see that R shows the row number of the first element of each row between straight brackets. This makes it easier to refer to a particular element, especially when the vectors are long. We can work with vectors in the same way as with single numbers. In principle all operations are carried out in an element wise fashion.

c(1, 2, 3) * c(4, 5, 6)

When we try to create a vector that consists of different data types they will be converted to a data type that is capable of containing all of them. For example:

c("eleven", 12)

The second element of the resulting vector is now also of type character. In general it is better not to trust this implicit conversion. Instead to it explicitly, in this case by using the function as.character().

An other way to create a vector is by using the function vector. vector('numeric', 8) creates a numeric vector of length 8. The vector function is often used to preallocate room where the results of future computations can be stored.

lists

Elements of a vector are always of the same type. A list differs from a vector by also allowing its elements to be of a different type. We can make a list using the function list.

list1 <- list("eleven", 12)
list2 <- list(c(1, 2, 3), c('foo', 'bar'))

We can also assign a name to the elements of a list:

list3 <- list(numbers=c(1, 2, 3), chars=c('foo', 'bar'))

It is also possible for a list to contain other lists.

list4 <- list(numbers=c(1, 2, 3), chars=c('aap', 'noot'), 
               sublist=list(1,'a'))

In many ways lists resemble data.frames. An important difference is that the length of all elements of a data.frame (that is the variables) must be the same. For a list this is not the case.

factors

A factoris a special kind of vector for categorical data. The factor contains different integers one for every category. Each unique value has an associated 'label' that tell us what the code means. Factors are frequently used when we model categorical data. An advantage of a factor over a character is that we can limit the number of possible outcomes. It is also less likely to make mistakes due to typing errors. Factors can be created by means of the function factor().

factor(c('male', 'female', 'male'))

When a factor is displayed R also shows us the unique values the variable can take. These are called the 'levels' of the factor.

Missing values

Whenever the value of a variable is missing this is denoted by NA in R. Usually this means that the values exists however we do not know it. Sometimes the result of a calculation is not finite (for example when we define a positive or negative number by zero). In this case the result is defined to be Inf of -Inf in R. When a value cannot be computed at all (for example when we divide zero by zero) R will define the result as NaN, which stands for 'Not a Number'. Finally, R sometimes uses the special value NULL to indicate that a variable is not yet defined. Here we will mostly deal with data that is just missing, that is NA.

stenw/BST02R documentation built on May 26, 2019, 4:35 a.m.