03 – Selection Syntax

Sebastian Raschka

Selection Syntax
- Selecting Elements from Vectors and Matrices (Single Bracket Indexing)
- Selecting from Data Frames

Source file: https://github.com/rasbt/R-notes/blob/master/03-selection-syntax.Rmd

Selection Syntax

Selecting Elements from Vectors and Matrices (Single Bracket Indexing)

We can use the single brackets [] indexing approach to return one or more elements from a vector or data frame, not that it returns the same object as the original.

vec_a <- c("x", "y", "z")

vec_a[1] # return a vector containing the first element

## [1] "x"

vec_a[1:2] # return a vector containing the first two elements

## [1] "x" "y"

We can also combine the single bracket selection approach with logical operators.
Suppose we have the following integer vector; we can use the >= operator to select all numbers that are equal to or larger than a certain value (here: greater than or equal to 5):

vec_b <- c(1:9)
vec_b

## [1] 1 2 3 4 5 6 7 8 9

vec_b[vec_b >= 5] 

## [1] 5 6 7 8 9

Also, we can chain multiple logical operators; let’s select all elements greater than 5 and smaller than 8:

vec_b[(vec_b > 5) & (vec_b < 8)]

## [1] 6 7

How about Matrices?

Analogous to the vector examples above, we can also use the single bracket notation to select elements and entire rows and columns from matrices. Let’s walk through some examples using a simple 3x2 matrix:

my_matrix <- matrix(1:6, 3, 2) # 3x2 matrix
my_matrix

##      [,1] [,2]
## [1,]    1    4
## [2,]    2    5
## [3,]    3    6

Use two values separated by a comma; the first value is the row index, the second value is the column index.
E.g., to select the 1st element in the 2nd row, we can do

my_matrix[2, 1]

## [1] 2

To select an entire row, we omit the value after the comma. E.g., to select the 2nd row:

my_matrix[2, ]

## [1] 2 5

Similarly, we can select an entire column by leaving the row value empty. E.g., to select the 2nd column:

my_matrix[, 2]

## [1] 4 5 6

We can also select with logical vectors and matrices. Suppose we have the following matrix with missing values:

my_matrix_with_na = my_matrix # make a copy

my_matrix_with_na[1, 2] <- NA # overwrite element with NA value
my_matrix_with_na

##      [,1] [,2]
## [1,]    1   NA
## [2,]    2    5
## [3,]    3    6

We can create a logical matrix, a mask, denoting wether a value is a NA or not:

mask <- is.na(my_matrix_with_na)
mask

##       [,1]  [,2]
## [1,] FALSE  TRUE
## [2,] FALSE FALSE
## [3,] FALSE FALSE

The reason why we call this mask is that it allows us to select us specific values from the original matrix, i.e., all the TRUE values:

my_matrix_with_na[mask]

## [1] NA

Note that we can also invert the selection using the ! operator (which is more useful in this case, assuming we want to filter out NA values). E.g., we can select all the non-NA values as follows:

!mask # inverted mask 

##      [,1]  [,2]
## [1,] TRUE FALSE
## [2,] TRUE  TRUE
## [3,] TRUE  TRUE

my_matrix_with_na[!mask]

## [1] 1 2 3 5 6

Selecting from Data Frames

Suppose we have the following data frame:

df <- data.frame(MyIntegerVar = 1:4, MyCharVar = c("A", "B", "A", "B"), MyBoolVar = c(T, F, T, T))
df

##   MyIntegerVar MyCharVar MyBoolVar
## 1            1         A      TRUE
## 2            2         B     FALSE
## 3            3         A      TRUE
## 4            4         B      TRUE

To get the 3rd row, we can use the same bracket notation as above for matrices:

df[3, ]

##   MyIntegerVar MyCharVar MyBoolVar
## 3            3         A      TRUE

Similarly, we can select the 3rd column as follows:

df[, 3] # returns a vector

## [1]  TRUE FALSE  TRUE  TRUE

# or

df[3] # returns a data frame with 1 column

##   MyBoolVar
## 1      TRUE
## 2     FALSE
## 3      TRUE
## 4      TRUE

Different Ways for Selecting Columns from a Data Frame

However, when working with data frames, we can take advantage of the column names when selecting columns. E.g., to select the 3rd column similar to df[3], we can do the following:

df['MyBoolVar']

##   MyBoolVar
## 1      TRUE
## 2     FALSE
## 3      TRUE
## 4      TRUE

And to select the third column similar to df[, 3] – returning a vector instead of a 1-column data frame – we can use the $ symbol as follows:

df$MyBoolVar

## [1]  TRUE FALSE  TRUE  TRUE

Selecting Elements from a Data Frame

Note that the selection syntax works similar to a matrix; to select the 1st element of the 2nd row, we can do

df[2, 1]

## [1] 2

however, we can also use the column name approach:

df[2, 'MyIntegerVar']

## [1] 2

df$MyIntegerVar[2]

## [1] 2

Conditional Selection

Also, similar to vectors and matrices, we can use conditional statements to select values. For instance, we can use the following to select all entries for which MyIntegerVar is greater than 1 and MyBoolVar is true

df[(df['MyIntegerVar'] > 1) & (df['MyBoolVar'] == TRUE), ] 

##   MyIntegerVar MyCharVar MyBoolVar
## 3            3         A      TRUE
## 4            4         B      TRUE