8 Subsetting rows
Selecting (subsetting) data by rows
This is another very common function we need to do on datasets, for example to remove incorrect data, or to work with only a portion of a dataset. If we want to subset rows by some variable in the data, we use another method, filter, which we’ll learn later.
Selecting rows of a dataset can be done with the verb slice()
. Unlike columns, rows generally do not have names, so we mostly select them numerically, by position. Similarly to select()
, slice()
first expects the dataframe, and after that the specific instructions on what to keep or remove. For example, to select the first and fifth rows of the titanic_df
dataframe:
Try it yourself: below, select the 20th and 50th rows of the dataframe.
As with select()
, we can choose a range using a colon (using glimpse here allows us to see the new size of the dataframe):
Try it yourself:
Select the first 50 rows, and everything except the 100th row. Functions often work in similar ways, especially if they are from the same suite (the Tidyverse), so you should be able to figure out how this works from our work on select()
.
slice_min and slice_max
There are also some helpful variations of slice()
. The two most common are slice_min()
and slice_max()
. These select a chosen number of rows based on some numerical variable (a column) in your data.
In this function, we use what are called ‘arguments’. These are further instructions, or parameters, which we pass to a function in order to specify what it should do. For slice_max()
, we have to tell it which column to use to get the maximum values with the argument order_by
, and how many rows to give with the argument n
.
For example, to select the ten youngest passengers in the dataset, we could use the following:
Another useful slice_
function is slice_sample()
, which picks a random sample of your data.
Try it yourself: Use a slice_ function to find the five oldest passengers (look at how we used select()
on the previous page to figure out how to do this).
slice_max()
works in the same way, but in reverse: