4. Wrangling data with dplyr

This worksheet provides a short interactive overview of the core verbs (functions) in dplyr to wrangle data. For more background and resources on data wrangling, see Resources: Wrangling data in the tidyverse.

Function	Description
select()	Subset columns.
filter()	Subset rows by conditions based on data in columns.
mutate()	Create new columns using information from other columns.
group_by()	Group the data by values in column such as by village or by sex.
summarize()	Aggregate the data, usually on grouped data, to create summary tables.
arrange()	Arrange the order of rows by values in columns. Often used with desc() to reverse default order.

We will use the penguins data for this demonstration and will make it into a tibble to get the nice tibble printout.

1 Subsetting

You can subset columns using select() to either explicitly keep columns or to discard columns using the not operator (!). To subset rows you can use filter(), which will keep rows based on one or more conditional statements. As we subset the data frame, keep your eye on the number of columns and rows of the resulting data frame at the top of the print out. The full penguins dataset has 8 columns, and 344 rows.

Try subsetting columns with select():

Try subsetting rows with filter():

2 Mutating

mutate() adds columns, usually based on the values in other columns. The argument structure within the function is the name of the new column equals the condition used to create the value. By default the condition is done by row and the new column is placed at the end of the data frame. Because of this, we can remove unnecessary columns to make it easier to see the new column and remove missing values using is.na().

Try mutate(): Create a new column named flipper_to_mass that calculates the ration of flipper length to body mass.

3 Summarizing

summarize() aggregates data, applying a summary function to a data frame that is usually grouped by values in a column created through group_by(). For instance, we might look at the average ratio of bill_dep to bill_len per penguin species and sex. Because we want to use sex as a group, we should probably remove any rows that have missing values for sex.

You will notice a message that tells us the resulting data frame still is still grouped by species. Look at the documentation for summarize() to see if you can quiet the message by altering the above code chunk and running it again.

Now make your own summary. Find the average weight of the penguins by species and sex.