One-hot encoding with pd.get_dummies

It is not uncommon in data analysis and machine learning applications to take data that is categorical in nature and convert it into a sequence of 0/1 values, as the latter can be more easily interpreted by numeric algorithms. This process is often called one-hot encoding, and the outputs are typically referred to as dummy indicators.

How to do it

Let’s start with a small pd.Series containing a discrete set of colors:

ser = pd.Series([
    "green",
    "brown",
    "blue",
    "amber",
    "hazel",
    "amber",
    "green",
    "blue",
    "green",
], name="eye_colors", dtype=pd.StringDtype())
ser

0    green
1    brown
2     blue
3    amber
4    hazel
5    amber
6    green
7     blue
8    green
Name: eye_colors, dtype: string

Passing this as an argument to pd.get_dummies will create a like-indexed pd.DataFrame with a...