9.7 Two-dimensional structures manipulation

9.7.1 Dimensions

• Get the number of rows and the number of columns:

# Create a data frame
d <- data.frame(c("Maria", "Juan", "Alba"),
    c(23, 25, 31),
    c(TRUE, TRUE, FALSE),
    stringsAsFactors = FALSE)
# number of rows
nrow(d)

## [1] 3

# number of columns
ncol(d)

## [1] 3

• Check the dimensions of the object: both number of rows and number of columns:

# first element: number of rows
# second element: number of columns
dim(d)

## [1] 3 3

• Dimension names

Column and/or row names can be added to matrices and data frames

colnames(d) <- c("Name", "Age", "Vegetarian")
rownames(d) <- c("Patient1", "Patient2", "Patient3")

Column and/or row names can be used to retrieve elements or sets of elements from a 2-dimensional object:

d[,"Name"]

## [1] "Maria" "Juan"  "Alba"

# same as:
d[,1]

## [1] "Maria" "Juan"  "Alba"

d["Patient3", "Age"]

## [1] 31

# same as:
d[3,2]

## [1] 31

# for data frames only, the $ sign can be used to retrieve columns:
# d$Name is d[,1] is d[, "Name"]

• Include names as you create objects:
  • Matrix:

m <- matrix(1:4, ncol=2, 
    dimnames=list(c("row1", "row2"), c("col1", "col2")))

+ Data frame:

df <- data.frame(col1=1:2, col2=1:2, 
    row.names=c("row1", "row2"))

9.7.2 Manipulation

Same principle as vectors… but in 2 dimensions!

Examples

• select the columns of b if at least one element in the 3rd row is less than or equal to 4:

# create b
b <- matrix(c(1, 0, 34, 44, 12, 4), 
    nrow=3, 
    ncol=2)
# third row of b:
b[3, ]

## [1] 34  4

# element(s) in the third row of b that is (are) less than or equal to 4
b[3, ] <= 4

## [1] FALSE  TRUE

# retrieve the corresponding sub-matrix
b[ ,b[3, ] <= 4]

## [1] 44 12  4

• Select rows of b if at least one element in column 2 is greater than 24:

# build data frame d
d <- data.frame(Name=c("Maria", "Juan", "Alba"), 
        Age=c(23, 25, 31),
        Vegetarian=c(TRUE, TRUE, FALSE),
        stringsAsFactors = FALSE)
rownames(d) <- c("Patient1", "Patient2", "Patient3")
# The following commands all output the same result:
d[d[,2] > 24, ]

##          Name Age Vegetarian
## Patient2 Juan  25       TRUE
## Patient3 Alba  31      FALSE

d[d[,"Age"] > 24, ]

##          Name Age Vegetarian
## Patient2 Juan  25       TRUE
## Patient3 Alba  31      FALSE

d[d$Age > 24, ]

##          Name Age Vegetarian
## Patient2 Juan  25       TRUE
## Patient3 Alba  31      FALSE

• Select patients (rows) based on 2 criteria: age of the patient (column 2) should be great than or equal to 25, and the patient should be vegetarian (column 3):

d[ d$Age >= 25 & d$Vegetarian == TRUE, ]

##          Name Age Vegetarian
## Patient2 Juan  25       TRUE

More useful commands

• Add a row or a column with rbind and cbind, respectively

# add a column
cbind(d, 1:3)

##           Name Age Vegetarian 1:3
## Patient1 Maria  23       TRUE   1
## Patient2  Juan  25       TRUE   2
## Patient3  Alba  31      FALSE   3

# add a row
rbind(d, 4:6)

##           Name Age Vegetarian
## Patient1 Maria  23          1
## Patient2  Juan  25          1
## Patient3  Alba  31          0
## 4            4   5          6

Add a patient to our data frame d:

d <- rbind(d, c("Jordi", 33, FALSE))

• Process the sum of all rows or all columns with rowSums and colSums, respectively.

# create a matrix
b <- matrix(1:20, ncol=4)
# process sum of rows and sum of cols
rowSums(b)

## [1] 34 38 42 46 50

colSums(b)

## [1] 15 40 65 90

• The apply function

Powerful tool to apply a command to all rows or all columns of a data frame or a matrix.
For example, instead of calculating the sum of each row, you might be interested in calculating the median ? But rowMedians doesn’t exist !

apply takes 3 arguments: - first argument X: 2-dimensional object - second argument MARGIN: apply by row or by column? + 1: by row + 2: by column - third argument FUN: function to apply to either rows or columns

# median value of each row of b
apply(X=b, MARGIN=1, FUN=median)

## [1]  8.5  9.5 10.5 11.5 12.5

# median value of each column of b
apply(X=b, MARGIN=2, FUN=median)

## [1]  3  8 13 18