- Review of last week's tasks
- Objects and assignments
- Fundamental data structures and operations: Vectors
- Fundamental data structures and operations: Data frames
November 01, 2018
Today's schedule
Review of last week's tasks
Solution for tasks #1
now online on https://wzbsocialsciencecenter.github.io/wzb_r_tutorial/
R Basics I
Objects and assignments
- an object is basically anything that can be named and assigned a value
x <- 2
- creates an object named
xand assigns the value2 <-is an assignment operator *- whole line is formally called a statement, it's evaluated once you press
ENTER - shortcut for
<-in RStudio:ALT+-
A little shortcut used in the slides:
(y <- x + 3)
## [1] 5
→ parentheses around assignment directly report the result underneath
* As in other programming languages, = can also be used for that. You can also reverse the assignment: 2 -> x. However, both is frowned upon. Please don't do it. The convention in R is to use <-.
Environment
- objects are kept in memory during the R session
- in RStudio, they are listed in the "Environment" tab in the top right pane
- use
rm(...)to delete objects from memory:
rm(x)
"Objects" vs. "Variables"
In Computer Science and programming, such objects are often called "variables", because they can change their value while a program or script is running.
This is quite different from what variables mean in Social Sciences. Please don't confuse this.
Naming things
You can name objects as you like. However, there are some rules:
Identifiers consist of a sequence of letters, digits, the period (‘.’) and the underscore. They must not start with a digit or an underscore, or with a period followed by a digit.
— R Language Definition
So this is fine:
another_one <- 1 another_2 <- 2 my.object.2... <- 123
Naming things
You can name objects as you like. However, there are some rules:
Identifiers consist of a sequence of letters, digits, the period (‘.’) and the underscore. They must not start with a digit or an underscore, or with a period followed by a digit.
— R Language Definition
This is not:
_fail <- 1 ## Error: unexpected input in "_"
2fail_again <- 'testing...' ## Error: unexpected symbol in "2fail_again"
Naming things
Which of these are not valid object names?
coffee2gox__x__.__._.x.1.1coffeegüzel2go
Naming things
Which of these are not valid object names?
coffee2gox__x__.__._.x.1.1coffeegüzel2go
Making comments
Everything starting with a # will be ignored by R. This is used for making source code comments or temporarily disabling some code:
# comment above c <- a + b # comment beside # d <- c * b # I disabled this statement
When working with the console, that's not really useful. But we'll need it later when we'll write real R scripts.
Fundamental data structures and operations
Vectors
R's most fundamental data structure: the vector. It stores one-dimensional data of a single type.
A vector is created with c(...).
Four types of vectors that you'll need:
- Numeric vector:
c(3.2, 0, -8.6, 3e-2)* - Logical vector:
c(FALSE, TRUE, TRUE, FALSE, TRUE) - Character string vector:
c("she", "said", '"ok?"') - Vector of factors:
factor(c("B", "B", "A", "C", "B"))
Single-element vectors are called scalars:
x <- 3 is the same as x <- c(3)
* Special type of numeric vector – the integer vector: c(3L, 20L, -1L, 3L)
Arithmetical operations
All arithmetical operations on vectors are performed element-wise:
a <- c(1.2, 0, -0.6, 0.25) b <- c(2, 3, 10, 0) a + b
## [1] 3.20 3.00 9.40 0.25
a - b
## [1] -0.80 -3.00 -10.60 0.25
a * b
## [1] 2.4 0.0 -6.0 0.0
a / b
## [1] 0.60 0.00 -0.06 Inf
(note how the division by zero produced an infinite number)
Arithmetical operations
You can create more complex expressions:
a^2 + b^2
## [1] 5.4400 9.0000 100.3600 0.0625
a + b / 2
## [1] 2.20 1.50 4.40 0.25
(a + b) / 2
## [1] 1.600 1.500 4.700 0.125
Vector recycling
What if the sizes of the vectors do not match?
a
## [1] 1.20 0.00 -0.60 0.25
c <- c(2, 3, 10) a * c
## Warning in a * c: longer object length is not a multiple of shorter object ## length
## [1] 2.4 0.0 -6.0 0.5
A warning is issued, still both vectors are multiplied; the shorter vector c is "recycled":
- \(a_1 c_1 = 1.2 \cdot 2 = 2.4\)
- \(a_2 c_2 = 0 \cdot 3 = 0\)
- \(a_3 c_3 = -0.6 \cdot 10 = -6\)
- \(\mathbf{a_4 c_1 = 0.25 \cdot 2 = 0.5}\)
Vector recycling
No warning is issued, if the larger vector is a multiple of the smaller vector:
a
## [1] 1.20 0.00 -0.60 0.25
d <- c(2, 3) a * d
## [1] 2.40 0.00 -1.20 0.75
→ d is recycled to form c(2, 3, 2, 3)
Multiplication by scalar is also a form of vector "recycling":
e <- 100 a * e
## [1] 120 0 -60 25
→ e is recycled to form c(100, 100, 100, 100)
Side note: Working with RStudio
The command history
- all commands entered in the console are recorded to "command history"
- whole history visible in "History" tab on top right
Side note: Working with RStudio
Using the history in the console
- you can browse through the history in the Console using the
UPandDOWNkeys - you can also use
CTRL+Rto interactively search through the history - this saves a lot of typing, e.g. when you want to correct a previous command
Side note: Working with RStudio
Autocompletion
- auto-completion of commands suggests objects after writing a few letters
- use the
TABkey to activate auto-completion - this saves a lot of typing when you have long object names
my_long_object_name <- c(1, 2, 3)
- now just type my_ in the console and press TAB - what happens when you type me (not my) and then activate auto-completion?
Vector concatenation
The function c(...) can also be used to concatenate vectors:
a
## [1] 1.20 0.00 -0.60 0.25
b
## [1] 2 3 10 0
c(a, b)
## [1] 1.20 0.00 -0.60 0.25 2.00 3.00 10.00 0.00
Vector concatenation
Vectors are always flat, i.e. one-dimensional. Hence, nesting the c() function will concatenate several vectors:
c(1, 2, 3, c(4, 5))
## [1] 1 2 3 4 5
You can append (or prepend) data to an already defined vector by appending and then re-assigning it to the same object name:
a
## [1] 1.20 0.00 -0.60 0.25
a <- c(a, 1, 2, 3)
a
## [1] 1.20 0.00 -0.60 0.25 1.00 2.00 3.00
Logical vectors
Logical vectors or boolean vectors contain only binary logical values TRUE and FALSE*. Logical operations like AND, OR, XOR can be applied.
is_female <- c(FALSE, TRUE, TRUE, FALSE, TRUE) younger30 <- c(TRUE, TRUE, FALSE, FALSE, FALSE)
Logical AND:
is_female & younger30
## [1] FALSE TRUE FALSE FALSE FALSE
Logical OR:
is_female | younger30
## [1] TRUE TRUE TRUE FALSE TRUE
Use case: Select observations according to several criteria, i.e. "participant is female and younger than 30".
* R defines the abbreviations T / F for TRUE / FALSE. Please don't use them. You may confuse them with a variable/object T or F.
Character string vectors
Character strings are used to store textual data.
s <- c("hello", "string")
s
## [1] "hello" "string"
You can use either double (") or single (') quotation marks to denote a string.
t <- c("world", 'example')
There are many functions that operate on strings, for example paste:
paste(s, t)
## [1] "hello world" "string example"
Use cases:
- storing names, addresses, comments, IDs, etc.
- performing quantitative text analysis
Factors
Factors are used to store categorical variables. They can be initialized with numeric or string vectors:
group <- factor(c("B", "B", "A", "C", "B"))
group
## [1] B B A C B ## Levels: A B C
Internally, each value is mapped to an integer (a category "code").
as.integer(group)
## [1] 2 2 1 3 2
Factor levels
group
## [1] B B A C B ## Levels: A B C
Levels denote the available categories for a variable. levels(x) returns the levels of a factor object x:
levels(group)
## [1] "A" "B" "C"
Levels can be specified explicitely:
factor(c('yes', 'no', 'yes', 'yes'), levels = c('yes', 'no', 'maybe'))
## [1] yes no yes yes ## Levels: yes no maybe
Ordered factors
Ordinal variables can be stored as ordered factors with ordered(...)
consent <- ordered(c('low', 'low', 'very low', 'high'),
levels = c('very low', 'low', 'neutral', 'high', 'very high'))
consent
## [1] low low very low high ## Levels: very low < low < neutral < high < very high
levels(consent)
## [1] "very low" "low" "neutral" "high" "very high"
You can also use numeric values as factors:
score <- ordered(c(3, 1, 1, 5), levels = 1:5) score
## [1] 3 1 1 5 ## Levels: 1 < 2 < 3 < 4 < 5
Type conversion
Conversion is the explicit process of changing a (vector's) data type.
Explicit type conversion can be done with a family of as.<TYPE>(...) functions, e.g. to convert a factor to an integer vector:
consent
## [1] low low very low high ## Levels: very low < low < neutral < high < very high
as.integer(consent)
## [1] 2 2 1 4
Or a logical vector to an integer vector:
younger30
## [1] TRUE TRUE FALSE FALSE FALSE
as.integer(younger30)
## [1] 1 1 0 0 0
Type conversion
You'll often have to convert strings (e.g read from a file) to numbers. The following is a character string vector:
(values <- c('1.2', '-3', "0.001"))
## [1] "1.2" "-3" "0.001"
Arithmetic operations fail, because we can't multiply a string with a number:
values * 3 ## Error in values * 3 : non-numeric argument to binary operator
We have to convert the string vector to a numeric vector first:
as.numeric(values) * 3
## [1] 3.600 -9.000 0.003
Type coercion
Coercion is the implicit process of a (vector's) data type being changed.
c(0.2, TRUE, 4)
## [1] 0.2 1.0 4.0
All elements of an atomic vector must be the same type, so when you attempt to combine different types they will be coerced to the most flexible type. Types from least to most flexible are: logical, integer, double, and character.
— Wickham 2014: Advanced R
c(0.2, TRUE, 4, '3.1')
## [1] "0.2" "TRUE" "4" "3.1"
Type coercion
Coercion happens automatically, esp. when using mathematical functions, which can come in handy:
younger30
## [1] TRUE TRUE FALSE FALSE FALSE
sum(younger30)
## [1] 2
→ sum() here counts all occurrences of TRUE because TRUE becomes 1 and FALSE becomes 0 when coerced to a numeric type.
Missings
NA is a special value reserved for denoting missing values. They can occur in all types of vectors:
(age <- c(32, 68, NA, 55, 35, NA, 55, 56))
## [1] 32 68 NA 55 35 NA 55 56
(smoker <- c(TRUE, NA, NA, TRUE, FALSE, FALSE, FALSE, TRUE))
## [1] TRUE NA NA TRUE FALSE FALSE FALSE TRUE
is.na(...) indicates which elements are missing:
is.na(age)
## [1] FALSE FALSE TRUE FALSE FALSE TRUE FALSE FALSE
We can combine this with sum(...) to find out, how many elements are missing in a vector:
sum(is.na(age))
## [1] 2
Missings
All operations that involve a calculation with a NA value result in NA:
a <- c(1, NA, 3) b <- c(NA, 8, 9) a + b
## [1] NA NA 12
sum(a)
## [1] NA
mean(b)
## [1] NA
We'll later learn how to ignore NAs for some calculations.
Vector length
A vector's size or length can be found out with length:
smoker
## [1] TRUE NA NA TRUE FALSE FALSE FALSE TRUE
length(smoker)
## [1] 8
→ counts the number of elements in a vector including NA values
Data frames
Data frames are the most common data structure for storing tabular data. They consist of columns and rows. Think of the columns as variables and rows as observations.
Each column is a vector with a specific type. All columns (vectors) have the same length.
Creating data frames
There are multiple ways to create a data frame:
1. By passing vectors
(cities <- data.frame(city = c('Berlin', 'Paris', 'Madrid'),
pop = c(3711930, 2206488, 3141991),
area = c(892, 105, 604)))
## city pop area ## 1 Berlin 3711930 892 ## 2 Paris 2206488 105 ## 3 Madrid 3141991 604
2. By loading data from a file
Note that there are many formats to load data from (Excel, CSV, Stata, etc.)
(cities <- read.csv('02rbasics1-data/cities.csv', stringsAsFactors = FALSE))
## city pop area ## 1 Berlin 3711930 892 ## 2 Paris 2206488 105 ## 3 Madrid 3141991 604
Creating data frames
3. By conversion from another data type
consent # this is a factor
## [1] low low very low high ## Levels: very low < low < neutral < high < very high
as.data.frame(consent)
## consent ## 1 low ## 2 low ## 3 very low ## 4 high
4. By using a builtin dataset
data("airquality")
head(airquality, 2)
## Ozone Solar.R Wind Temp Month Day ## 1 41 190 7.4 67 5 1 ## 2 36 118 8.0 72 5 2
Data frame inspection
cities
## city pop area ## 1 Berlin 3711930 892 ## 2 Paris 2206488 105 ## 3 Madrid 3141991 604
Finding out the number of columns:
ncol(cities)
## [1] 3
Finding out the number of rows:
nrow(cities)
## [1] 3
Data frame inspection
Getting/setting the column names:
colnames(cities)
## [1] "city" "pop" "area"
colnames(cities) <- c('city', 'population_cityarea', 'area_km2')
cities
## city population_cityarea area_km2 ## 1 Berlin 3711930 892 ## 2 Paris 2206488 105 ## 3 Madrid 3141991 604
Data frame inspection
Getting/setting the row names:
rownames(cities)
## [1] "1" "2" "3"
rownames(cities) <- c('BLN', 'PRS', 'MDRD')
cities
## city population_cityarea area_km2 ## BLN Berlin 3711930 892 ## PRS Paris 2206488 105 ## MDRD Madrid 3141991 604
Setting row names is usually not necessary as you can use a column as row identifier.
Accessing columns in data frames
cities
## city population_cityarea area_km2 ## BLN Berlin 3711930 892 ## PRS Paris 2206488 105 ## MDRD Madrid 3141991 604
Remember: Each data frame column is a vector. You can access a column and thereby obtain its data vector using the $ sign:
cities$population_cityarea
## [1] 3711930 2206488 3141991
You can work with them as usual:
max(cities$population_cityarea)
## [1] 3711930
How would you calculate the mean of the area in the cities data frame?
Column names
- column name rules are more relaxed than object name rules
- e.g. you can use spaces but you have to use a special syntax for access then:
cities$`city name` # assuming that I renamed column "city" to "city name" ## [1] "Berlin" "Paris" "Madrid"
- it's generally a good idea to avoid spaces and other special characters also in column names - arkward column names can happen during data import → you should rename the columns before further processing!
Adding and removing columns
How would you calculate the population density (number of inhabitants per km²)?
New columns can be added by assigning a vector to a new column name:
cities$pop_density <- cities$population_cityarea / cities$area_km2 cities
## city population_cityarea area_km2 pop_density ## BLN Berlin 3711930 892 4161.357 ## PRS Paris 2206488 105 21014.171 ## MDRD Madrid 3141991 604 5201.972
There are many ways to drop a column, one is to assign the special value NULL:
cities$pop_density <- NULL cities
## city population_cityarea area_km2 ## BLN Berlin 3711930 892 ## PRS Paris 2206488 105 ## MDRD Madrid 3141991 604
Accessing rows/observations
Accessing rows (i.e. observations) is called subsetting or filtering. Subsetting does not yield a vector but another data frame (a subset of the source data frame).
You will learn how to do this with a package called dplyr in one of the next sessions.
Tasks
Notes on the next three weeks
There won't be tutorials at the WZB for the next three weeks.
There is material for self-study at:
These documents cover the following topics:
R Basics II: Functions, object types, comparisons and logical expressions
R Basics III: Subsetting, working with R scripts, reading/writing data from/to file
Transforming data with R I: Filtering and ordering observations, selecting variables, adding new variables
Work through these documents and tasks, because they are essential for all the other sessions.
Lena Hipp will provide consultation hours.
Tasks
- Install SWIRL courses that we'll need for the next sessions. You should already have installed the package swirl. If not, you need to do that first (see previous session's slides). Next, install the interactive R courses as described on the next slide.
- Complete lessons 1 and 4 of SWIRL Course "R Programming". Lesson 3 is optional, lesson 2 should be skipped.
- Take a look at the last 5 tweets of WZB_Berlin
- Create two vectors
retweetsandlikesthat contain the respective numbers from the last 5 tweets - Create a third vector
tweet_idsthat contains the letters a to e as identifiers for the tweets - Check the data type of all three vectors using the function
class(...) - Look at 5 more tweets, append the respective data to the vectors
- Create a dataframe
tweetstatsfrom the three vectors - Add an additional variable/column to
tweetstatsnamedinteractionswhich is the sum of retweets and likes for each observation
- Create two vectors
(Continued on next slide)
Tasks
-
As in the previous session's tasks, we'll work with the
catsdataset from the package MASS. - Load the package and the dataset.
- How do you bring up the dataset documentation / help for the dataset?
- Identify the number of rows and columns in the dataset by using the respective R functions.
- Identify the column names using the respective R function.
-
What are the data types of the columns in the dataset? Again, use
class(...)to answer this question. - What if you recorded two more variables: Age and whether the cat has heart problems. Which data types would you choose for each variable?
-
Create a new column
wt_ratiowhich is the ratio of heart and body weight. Make sure to bring both variables to a common unit of measurement (i.e. both in grams or kilograms).
Notes on installing SWIRL courses
First, load the package:
library(swirl)
If an error message shows up, then you probably didn't install the package. You need to do this first with the command install.packages('swirl').
Next, install three courses with the follwing commands (see next slide, in case one of this commands fails):
install_course_github("swirldev", "R Programming")
install_course_github("swirldev", "Getting and Cleaning Data")
install_course_github("swirldev", "Exploratory Data Analysis")
You can start SWIRL with the following command which will guide you through the first steps:
swirl()
One additional note: When you receive the question "Would you like to receive credit for completing this course on Coursera.org?", select "No".
Notes on installing SWIRL courses
In case installing a SWIRL course fails, do the follwing:
- Download the ZIP file
swirl_courses-master.zipfrom the following address: https://github.com/swirldev/swirl_courses/archive/master.zip - Move this file to your current working directory (you can find that out via
getwd()). - Execute the following commands:
install_course_zip("swirl_courses-master.zip", multi=TRUE,
which_course = "R Programming")
install_course_zip("swirl_courses-master.zip", multi=TRUE,
which_course = "Getting and Cleaning Data")
install_course_zip("swirl_courses-master.zip", multi=TRUE,
which_course = "Exploratory Data Analysis")