Introduction to data frames Steve Bagley somgen223.stanford.edu 1 - - PowerPoint PPT Presentation

Introduction to data frames

Steve Bagley

somgen223.stanford.edu 1

Using packages from the tidyverse

somgen223.stanford.edu 2

Need to install the tidyverse set of packages

• Type this:

install.packages("tidyverse")

• “tidyverse” is a coherent set of packages for operating a kind of data called the

“data frame.”

• It is not built-in, so you need to install it (once), then load it each time you

restart R.

• Put library(tidyverse) or library("tidyverse") at the top of every

script file.

• More about packages later.

somgen223.stanford.edu 3

Data frame: a two-dimensional data structure

A data frame is one of the most powerful features in R.

• It is a rectangular data structure that can contain different types of data, similar

to an Excel spreadsheet.

• Typically, each row in a data frame contains information about one instance of

some (real-world) object.

• Each column can be thought of as a variable, containing the values for the

corresponding instances.

• All the values in one column should be of the same type, but different columns

can be of different types.

somgen223.stanford.edu 4

Data frame example

bod <- as_tibble(BOD) bod # A tibble: 6 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3 3 3 19 4 4 16 5 5 15.6 6 7 19.8

• This data set contains data on biological oxygen demand.
• A tibble is a kind of data frame. This one has 6 rows and 2 columns.
• Across the top is the name of each column. The next row shows the type of data

in the column. <dbl>, means double-precision floating point number.

• The row numbers are added for printing. They are not part of the data frame.

somgen223.stanford.edu 5

Data frame example

bod # A tibble: 6 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3 3 3 19 4 4 16 5 5 15.6 6 7 19.8 Check claim on earlier slide:

• Typically, each row in a data frame describes an instance of some (real-world)
• bject. (Yes: one row for each subject, time, or measurement.)
• Each column contains the values of a variable for the corresponding instance.

(Yes: one column for each variable.)

somgen223.stanford.edu 6

Data frame functions

• The rest of this section shows data frame functions (“verbs”).
• Each function takes a data frame and produces a new data frame.

somgen223.stanford.edu 7

Data frame function: filtering rows

filter(bod, Time < 4) # A tibble: 3 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3 3 3 19

• This produces (and prints out) a new tibble, which contains all the rows where

the Time value in that row is less than 4.

• There are only 3 rows in this data frame.
• There are still 2 columns.
• filter does not modify bod, which still has 6 rows and 2 columns.

somgen223.stanford.edu 8

Combining constraints in filter

filter(bod, Time < 4, demand <= 16) # A tibble: 2 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3

• This filters by the conjunction of the two constraints—both must be satisfied.
• Constraints appear as second (and third, …) arguments, separated by commas.

somgen223.stanford.edu 9

Disjunction with filter

filter(bod, Time < 4 | demand <= 16) # A tibble: 5 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3 3 3 19 4 4 16 5 5 15.6

• This filters by the disjunction of the two constraints—either must be satisfied.
• More about logical operators later.

somgen223.stanford.edu 10

Filtering out all rows

filter(bod, Time > 10) # A tibble: 0 x 2 # ... with 2 variables: Time <dbl>, demand <dbl>

• If the constraint is too severe, then you will produce a tibble with zero rows.

somgen223.stanford.edu 11

Data frame function: select columns

select(bod, demand) # A tibble: 6 x 1 demand <dbl> 1 8.3 2 10.3 3 19 4 16 5 15.6 6 19.8

• The select function will return a subset of the tibble, using only the requested

columns in the order specified.

somgen223.stanford.edu 12

Use - to leave out a column

select(bod, -Time) # A tibble: 6 x 1 demand <dbl> 1 8.3 2 10.3 3 19 4 16 5 15.6 6 19.8

• The - operator can be used to leave out a column.

somgen223.stanford.edu 13

Data frame function: arrange to sort rows

arrange(bod, demand) # A tibble: 6 x 2 Time demand <dbl> <dbl> 1 1 8.3 2 2 10.3 3 5 15.6 4 4 16 5 3 19 6 7 19.8

• arrange takes a data frame and a column, and sorts the rows by the values in

that column in ascending order.

somgen223.stanford.edu 14

Use desc to sort from high to low

arrange(bod, desc(demand)) # A tibble: 6 x 2 Time demand <dbl> <dbl> 1 7 19.8 2 3 19 3 4 16 4 5 15.6 5 2 10.3 6 1 8.3

somgen223.stanford.edu 15

Interlude

somgen223.stanford.edu 16

Arguments by position

1:5 [1] 1 2 3 4 5 5:1 [1] 5 4 3 2 1 seq(1, 5) [1] 1 2 3 4 5 seq(5, 1) [1] 5 4 3 2 1

• seq is the function equivalent of the colon operator.
• Arguments can be specified by position, with one supplied argument for each

name in the function parameter list, and in the same order.

• Arguments are separated by commas.

somgen223.stanford.edu 17

Arguments by name

seq(from = 1, to = 5) [1] 1 2 3 4 5 seq(to = 5, from = 1) [1] 1 2 3 4 5

• Arguments can be supplied by name using the syntax, variable = value.
• When using names, the order of the named arguments does not matter.
• You can mix positional and named arguments (carefully).
• Do not use <- in place of = when specifying named arguments.

somgen223.stanford.edu 18

Using the correct argument name

seq(1, 5) [1] 1 2 3 4 5 seq(from = 1, to = 5) [1] 1 2 3 4 5 seq(begin = 1, end = 5) Warning: In seq.default(begin = 1, end = 5) : extra arguments 'begin', 'end' will be disregarded [1] 1

• You have to use the correct argument name.

somgen223.stanford.edu 19

How to find the names of a function’s arguments

• How can you figure out the names of seq’s arguments?
• Answer: the arguments are listed in the R documentation of the function.

## Try this: ?seq

somgen223.stanford.edu 20

How to assign and evaluate in one line

x <- 10 x [1] 10 (x <- 10) [1] 10

• When you are typing to R, it is very common to want to assign a variable and

return the value of the assignment.

• But ordinarily, R does not return a visible result from an assignment expression.
• You can force it to do so in a single expression by putting parentheses, ( ),

around the assignment.

• This is used throughout the R documentation and some of the slides for this

course, so you should be familiar with this bit of syntax.

somgen223.stanford.edu 21

Back to data frame functions

somgen223.stanford.edu 22

Data frame function: mutate to compute new column

(bod2 <- mutate(bod, inv_demand = 1 / demand)) # A tibble: 6 x 3 Time demand inv_demand <dbl> <dbl> <dbl> 1 1 8.3 0.120 2 2 10.3 0.0971 3 3 19 0.0526 4 4 16 0.0625 5 5 15.6 0.0641 6 7 19.8 0.0505

• This uses mutate to add a new column to bod which is the reciprocal of demand.
• Note use of = to assign to new column. Do not use <- here!
• The result of this is a new data frame with the new column. You must assign to a

new (or old) name to save the result.

• Note use of ( ) to assign and evaluate on a single line.

somgen223.stanford.edu 23

Exercise: change units in orange trees

• range <- as_tibble(Orange)

head(orange) # A tibble: 6 x 3 Tree age circumference <ord> <dbl> <dbl> 1 1 118 30 2 1 484 58 3 1 664 87 4 1 1004 115 5 1 1231 120 6 1 1372 142

• Add a new column to orange called circum_in which is the circumference in

inches, not in millimeters.

• Hint: 1 in = 2.54 cm

somgen223.stanford.edu 24

Answer: change units in orange trees

• range <- mutate(orange, circum_in = circumference/(10 * 2.54))
• range

# A tibble: 35 x 4 Tree age circumference circum_in <ord> <dbl> <dbl> <dbl> 1 1 118 30 1.18 2 1 484 58 2.28 3 1 664 87 3.43 4 1 1004 115 4.53 5 1 1231 120 4.72 6 1 1372 142 5.59 7 1 1582 145 5.71 8 2 118 33 1.30 9 2 484 69 2.72 10 2 664 111 4.37 # ... with 25 more rows

somgen223.stanford.edu 25

Use read_csv to read csv file (from file or url)

cw <- read_csv("https://somgen223.stanford.edu/data/cw.csv")

• We will need that directory many times in this course, so use str_c to construct

the location for read_csv data_dir <- "https://somgen223.stanford.edu/data/" cw <- read_csv(str_c(data_dir, "cw.csv"))

• data_dir holds the first part of the url as a string
• str_c glues all of its arguments together into a single string, like so:

str_c("abc", "def") [1] "abcdef"

somgen223.stanford.edu 26

Reading

• Read: 5 Data transformation | R for Data Science (sections 5.1 to 5.5)

somgen223.stanford.edu 27