Dummy variables (or binary variables) are commonly used in statistical analyses and in more simple descriptive statistics. A dummy column is one which has a value of one when a categorical event occurs and a zero when it doesn’t occur. In most cases this is a feature of the event/person/object being described. For example, if the dummy variable was for occupation being an R programmer, you can ask, “is this person an R programmer?” When the answer is yes, they get a value of 1, when it is no, they get a value of 0.
We’ll start with a simple example and then go into using the function
dummy_cols(). You can also use the function
dummy_columns() which is identical to
Imagine you have a data set about animals in a local shelter. One of the columns in your data is what animal it is: dog or cat.
To make dummy columns from this data, you would need to produce two new columns. One would indicate if the animal is a dog, and the other would indicate if the animal is a cat. Each row would get a value of 1 in the column indicating which animal they are, and 0 in the other column.
In the function dummy_cols, the names of these new columns are concatenated to the original column and separated by an underscore.
With an example like this, it is fairly easy to make the dummy
dummy_cols() automates the process, and
is useful when you have many columns to general dummy variables from or
with many categories within the column.
The object fastDummies_example has two character
type columns, one integer column, and a Date column. By default,
dummy_cols() will make dummy variables from factor or
character columns only. This is because in most cases those are the only
types of data you want dummy variables from. If those are the only
columns you want, then the function takes your data set as the first
parameter and returns a data.frame with the newly created variables
appended to the end of the original data.
In some situations, you would want columns with types other than factor and character to generate dummy variables. For example, a column of years would be numeric but could be well-suited for making into dummy variables depending on your analysis. Use the select_columns parameter to select specific columns to make dummy variables from.
The final option for
remove_first_dummy which by default is FALSE. If TRUE, it
removes the first dummy variable created from each column. This is done
to avoid multicollinearity in a multiple regression model caused by
included all dummy variables. The “first” dummy variable is the one at
the top of the rows (i.e. the first value that is not NA).