We showed one way to read data into R using read.csv. In this module, we will show you how to:

1. Select specific elements of an object by an index or logical condition
2. Renaming columns of a data.frame
3. Subset rows of a data.frame
4. Subset columns of a data.frame
5. Add/remove new columns to a data.frame
6. Order the columns of a data.frame
7. Order the rows of a data.frame

We will show you how to do each operation in base R then show you how to use the dplyr package to do the same operation (if applicable).

Many resources on how to use dplyr exist and are straightforward:

• https://cran.rstudio.com/web/packages/dplyr/vignettes/
• https://stat545-ubc.github.io/block009_dplyr-intro.html
• https://www.datacamp.com/courses/dplyr-data-manipulation-r-tutorial

Often you only want to look at subsets of a data set at any given time. As a review, elements of an R object are selected using the brackets ([ and ]).

For example, x is a vector of numbers and we can select the second element of x using the brackets and an index (2):

x = c(1, 4, 2, 8, 10)
x[2]

[1] 4

We can select the fifth or second AND fifth elements below:

x = c(1, 2, 4, 8, 10)
x[5]

[1] 10

x[c(2,5)]

[1]  2 10

You can put a minus (-) before integers inside brackets to remove these indices from the data.

x[-2] # all but the second

[1]  1  4  8 10

Note that you have to be careful with this syntax when dropping more than 1 element:

x[-c(1,2,3)] # drop first 3

[1]  8 10

# x[-1:3] # shorthand. R sees as -1 to 3
x[-(1:3)] # needs parentheses

[1]  8 10

What about selecting rows based on the values of two variables? We use logical statements. Here we select only elements of x greater than 2:

x

[1]  1  2  4  8 10

x > 2

[1] FALSE FALSE  TRUE  TRUE  TRUE

x[ x > 2 ]

[1]  4  8 10

You can have multiple logical conditions using the following:

• & : AND
• | : OR

x[ x > 2 & x < 5 ]

[1] 4

x[ x > 5 | x == 2 ]

[1]  2  8 10

The which functions takes in logical vectors and returns the index for the elements where the logical value is TRUE.

which(x > 5 | x == 2) # returns index

[1] 2 4 5

x[ which(x > 5 | x == 2) ]

[1]  2  8 10

x[ x > 5 | x == 2 ]

[1]  2  8 10

Here we create a toy data.frame named df using random data:

set.seed(2016) # reproducbility
df = data.frame(x = c(1, 2, 4, 10, 10),
                x2 = rpois(5, 10),
                y = rnorm(5),
                z = rpois(5, 6)
                )

We can use the colnames function to directly reassign column names of df:

colnames(df) = c("x", "X", "y", "z")
head(df)

   x  X          y  z
1  1  7 -0.2707606  6
2  2  6 -1.1179372  4
3  4 10 -1.3473558  7
4 10 13  0.4832675 10
5 10 13  0.1523950  5

colnames(df) = c("x", "x2", "y", "z") #reset

We can assign the column names, change the ones we want, and then re-assign the column names:

cn = colnames(df)
cn[ cn == "x2"] = "X"
colnames(df) = cn
head(df)

   x  X          y  z
1  1  7 -0.2707606  6
2  2  6 -1.1179372  4
3  4 10 -1.3473558  7
4 10 13  0.4832675 10
5 10 13  0.1523950  5

colnames(df) = c("x", "x2", "y", "z") #reset

library(dplyr)

Attaching package: 'dplyr'

The following objects are masked from 'package:stats':

    filter, lag

The following objects are masked from 'package:base':

    intersect, setdiff, setequal, union

Note, when loading dplyr, it says objects can be "masked". That means if you use a function defined in 2 places, it uses the one that is loaded in last.

For example, if we print filter, then we see at the bottom namespace:dplyr, which means when you type filter, it will use the one from the dplyr package.

filter

function (.data, ...) 
{
    filter_(.data, .dots = lazyeval::lazy_dots(...))
}
<environment: namespace:dplyr>

A filter function exists by default in the stats package, however. If you want to make sure you use that one, you use PackageName::Function with the colon-colon ("::") operator.

head(stats::filter,2)

                                                               
1 function (x, filter, method = c("convolution", "recursive"), 
2     sides = 2L, circular = FALSE, init = NULL)               

This is important when loading many packages, and you may have some conflicts/masking:

To rename columns in dplyr, you use the rename command

df = dplyr::rename(df, X = x2)
head(df)

   x  X          y  z
1  1  7 -0.2707606  6
2  2  6 -1.1179372  4
3  4 10 -1.3473558  7
4 10 13  0.4832675 10
5 10 13  0.1523950  5

df = dplyr::rename(df, x2 = X) # reset

We can grab the x column using the $ operator.

df$x

[1]  1  2  4 10 10

We can also subset a data.frame using the bracket [, ] subsetting.

For data.frames and matrices (2-dimensional objects), the brackets are [rows, columns] subsetting. We can grab the x column using the index of the column or the column name ("x")

df[, 1]

[1]  1  2  4 10 10

df[, "x"]

[1]  1  2  4 10 10

We can select multiple columns using multiple column names:

df[, c("x", "y")]

   x          y
1  1 -0.2707606
2  2 -1.1179372
3  4 -1.3473558
4 10  0.4832675
5 10  0.1523950

The select command from dplyr allows you to subset

select(df, x)

   x
1  1
2  2
3  4
4 10
5 10

The select command from dplyr allows you to subset columns of

select(df, x, x2)

   x x2
1  1  7
2  2  6
3  4 10
4 10 13
5 10 13

select(df, starts_with("x"))

   x x2
1  1  7
2  2  6
3  4 10
4 10 13
5 10 13

Let's select rows 1 and 3 from df using brackets:

df[ c(1, 3), ]

  x x2          y z
1 1  7 -0.2707606 6
3 4 10 -1.3473558 7

Let's select the rows of df where the x column is greater than 5 or is equal to 2. Without any index for columns, all columns are returned:

df[ df$x > 5 | df$x == 2, ]

   x x2          y  z
2  2  6 -1.1179372  4
4 10 13  0.4832675 10
5 10 13  0.1523950  5

We can subset both rows and colums at the same time:

df[ df$x > 5 | df$x == 2, c("y", "z")]

           y  z
2 -1.1179372  4
4  0.4832675 10
5  0.1523950  5

The command in dplyr for subsetting rows is filter. Try ?filter

filter(df, x > 5 | x == 2)

   x x2          y  z
1  2  6 -1.1179372  4
2 10 13  0.4832675 10
3 10 13  0.1523950  5

Note, no $ or subsetting is necessary. R "knows" x refers to a column of df.

By default, you can separate conditions by commas, and filter assumes these statements are joined by &

filter(df, x > 2 & y < 0)

  x x2         y z
1 4 10 -1.347356 7

filter(df, x > 2, y < 0)

  x x2         y z
1 4 10 -1.347356 7

You can combine filter and select to subset the rows and columns, respectively, of a data.frame:

select(filter(df, x > 2 & y < 0), y, z)

          y z
1 -1.347356 7

In R, the common way to perform multiple operations is to wrap functions around each other in a nested way such as above

One can also create temporary objects and reassign them:

df2 = filter(df, x > 2 & y < 0)
df2 = select(df2, y, z)

There is another (newer) way of performing these operations, called "piping". It is becoming more popular as it's easier to read:

df %>% filter(x > 2 & y < 0) %>% select(y, z)

          y z
1 -1.347356 7

It is read: "take df, then filter the rows and then select y, z".

You can add a new column, called newcol to df, using the $ operator:

df$newcol = 5:1
df$newcol = df$x + 2

You can remove a column by assigning to NULL:

df$newcol = NULL

or selecing only the columns that were not newcol:

df = df[, colnames(df) != "newcol"]

You can also "column bind" a data.frame with a vector (or series of vectors), using the cbind command:

cbind(df, newcol = 5:1)

   x x2          y  z newcol
1  1  7 -0.2707606  6      5
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      3
4 10 13  0.4832675 10      2
5 10 13  0.1523950  5      1

The mutate function in dplyr allows you to add or replace columns of a data.frame:

mutate(df, newcol = 5:1)

   x x2          y  z newcol
1  1  7 -0.2707606  6      5
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      3
4 10 13  0.4832675 10      2
5 10 13  0.1523950  5      1

print({df = mutate(df, newcol = x + 2)})

   x x2          y  z newcol
1  1  7 -0.2707606  6      3
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      6
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12

The NULL method is still very common.

The select function can remove a column with a minus (-), much like removing rows:

select(df, -newcol)

   x x2          y  z
1  1  7 -0.2707606  6
2  2  6 -1.1179372  4
3  4 10 -1.3473558  7
4 10 13  0.4832675 10
5 10 13  0.1523950  5

Remove newcol and y

select(df, -one_of("newcol", "y"))

   x x2  z
1  1  7  6
2  2  6  4
3  4 10  7
4 10 13 10
5 10 13  5

We can use the colnames function to get the column names of df and then put newcol first by subsetting df using brackets:

cn = colnames(df)
df[, c("newcol", cn[cn != "newcol"]) ]

  newcol  x x2          y  z
1      3  1  7 -0.2707606  6
2      4  2  6 -1.1179372  4
3      6  4 10 -1.3473558  7
4     12 10 13  0.4832675 10
5     12 10 13  0.1523950  5

The select function can reorder columns. Put newcol first, then select the rest of columns:

select(df, newcol, everything())

  newcol  x x2          y  z
1      3  1  7 -0.2707606  6
2      4  2  6 -1.1179372  4
3      6  4 10 -1.3473558  7
4     12 10 13  0.4832675 10
5     12 10 13  0.1523950  5

We use the order function on a vector or set of vectors, in increasing order:

df[ order(df$x), ]

   x x2          y  z newcol
1  1  7 -0.2707606  6      3
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      6
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12

The decreasing argument will order it in decreasing order:

df[ order(df$x, decreasing = TRUE), ]

   x x2          y  z newcol
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12
3  4 10 -1.3473558  7      6
2  2  6 -1.1179372  4      4
1  1  7 -0.2707606  6      3

You can pass multiple vectors, and must use the negative (using -) to mix decreasing and increasing orderings (sort increasing on x and decreasing on y):

df[ order(df$x, -df$y), ]

   x x2          y  z newcol
1  1  7 -0.2707606  6      3
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      6
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12

The arrange function can reorder rows By default, arrange orders in ascending order:

arrange(df, x)

   x x2          y  z newcol
1  1  7 -0.2707606  6      3
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      6
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12

Use the desc to arrange the rows in descending order:

arrange(df, desc(x))

   x x2          y  z newcol
1 10 13  0.4832675 10     12
2 10 13  0.1523950  5     12
3  4 10 -1.3473558  7      6
4  2  6 -1.1179372  4      4
5  1  7 -0.2707606  6      3

It is a bit more straightforward to mix increasing and decreasing orderings:

arrange(df, x, desc(y))

   x x2          y  z newcol
1  1  7 -0.2707606  6      3
2  2  6 -1.1179372  4      4
3  4 10 -1.3473558  7      6
4 10 13  0.4832675 10     12
5 10 13  0.1523950  5     12

The transmute function in dplyr combines both the mutate and select functions. One can create new columns and keep the only the columns wanted:

transmute(df, newcol2 = x * 3, x, y)

  newcol2  x          y
1       3  1 -0.2707606
2       6  2 -1.1179372
3      12  4 -1.3473558
4      30 10  0.4832675
5      30 10  0.1523950