Selecting parts of objects
MEB433 and MEB434
Lukáš Lehotský and Petr Ocelík
https://campus.datacamp.com/courses/free-
introduction-to-r
Functions: most useful basic functions
c() # combine two or more elements into an object
class() # explore elements’ data class
length() # explore number of first dim. of object
dim() # explore dimensions of two-dimensional obj.
nrow() # number of rows
ncol() # number of columns
head() # first few rows of data
tail() # last few rows of data
str() # explore structure of object
names() # names in the named vector - one dimension
rownames() # names of rows - two dimensions
colnames() # names of columns - two dimensions
Selecting: logic of indexes
• Important to realize whole logic of R is relying on position of
objects
• Any object consists of N elements (N = 1, 2, … , n)
• Any element/nested object might be selected based on its
position (also called index)
• Selection has to include all dimensions (always)
object[dim 1,...,dim n]
Selecting: one dimension
peaches <- c("first","second","third","fourth")
1st
2nd
3rd
4th
Selecting: one dimension
peaches <- c("first","second","third","fourth")
peaches[3]
[1] "third"
1st 2nd 3rd 4th
Selecting: two dimensions
• Data frame/matrix
• Selection of rows and/or columns
• Always have to include both
dimensions
• If dimension is empty, returns all values
– think of it as a restriction on certain
dimmension
object[ row , column ]
Selecting: two dimensions
rack[ 2 , ]
Selecting: two dimensions
rack[ 2 , ]
2nd row
Selecting: two dimensions
rack[ 2 , ]
2nd row
all columns
Selecting: two dimensions
rack[ 2 , ]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
rack["redapples", ]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
rack["redapples", ]
rack["greenapples",4]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
rack["redapples", ]
rack["greenapples",4]
Selecting: two dimensions
rack[ 2 , ]
rack[ , 2]
rack["redapples", ]
rack["greenapples",4]
Selecting: two dimensions
• Data frame/matrix df with car information below
• Selection of rows and/or columns
• Always have to include both dimensions
• If dimension is empty, returns all values
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
Selecting: two dimensions, single row
• Output of row-wise selection is always data frame
df[ 2 , ]
cars type price consumption
2 Audi A4 1164000 5.9
Selecting: two dimensions, single column
• Output of column-wise selection
• Is a vector if one column is selected
• Is a data frame if two or more columns selected
df[ , 1 ]
[1] "BMW" "Audi" "VW"
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
Selecting: two dimensions, row and column
• Output of selection on both dimensions
• Is vector if one column is selected
• Is data frame otherwise
df[ 2 , 1 ]
[1] "Audi"
Selecting: two dimensions, column by name
• If columns have names, there is way of selecting column by
its name
• Column names have to be defined (check with colnames())
• Column name should be in parentheses and in place of column
dimension
df[ , "cars" ]
[1] "BMW" "Audi" "VW"
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
Selecting: two dimensions, column by name
• Selection of one column returns vector
• The transformation might be undesirable
• Additional argument drop=FALSE
• Should be included after the last dimension
• Preserves the object type as data frame
df[ , "cars" , drop=FALSE ]
cars
1 "BMW"
2 "Audi"
3 "VW"
Selecting: two dim, multiple rows/columns
• Selection of multiple columns/rows possible through nesting
– using functions to construct the index
• Most commonly, function c() will allow to select more than one
column or row (e.g. c(1, 3))
• Uninterrupted range of columns can be selected using colon sign “:”
(e.g. 1:100)
df[ , c(1 , 3) ]
cars type price consumption
1 BMW 3 1200000 6.2
2 Audi A4 1164000 5.9
3 VW Passat 950500 5.9
Selecting: two dimensions, direct col. name
• A $ sign allows to call variable by its name directly
• Applies to data frame only, allows to select only one column
• No square brackets needed
• Output is one-dimensional vector only (can’t preserve the data
frame structure)
df$cars
[1] "BMW" "Audi" "VW"
df$price
[1] 1200000 1164000 950500
class(df$price)
[1] "numeric"
Selecting: operators
Operator Description
[ [[ Selection by index
$ @ Selection by variable/object name
: Sequence (from:to) operator
Source: Adler, J. (2012). R in a nutshell. Pp. 86.
Practice 1
• Create an object from the R dataset “eurodist”
• Explore the data source and transform it to the most suitable
format for data manipulation (matrix)
• Extract distances from Stockholm into separate vector
• Find shortest and longest distance from Stockholm (google
appropriate function)
• Figure out what is the distance between Cologne and Lisbon
Practice 2
• Install and load the package “poliscidata”
• Create a new object in your environment using the dataset
“world”
• Explore variables in the dataset (use colnames() function)
• Extract countries into separate vector
• Extract country IDs into separate data frame
• Find the Czech Republic in the data frame
• Extract the appropriate row from the data frame
• Find all V4 countries (CZ, SK, HU, PL)
• Extract them into a separate data set
• Extract only freedom indicators for these countries (all column names
starting with “free_”)