Imagine you have a dataframe with a column containing categorical variables. For example,
>>> df
num_1 num_2 cat
0 1 2 a
1 4 5 b
2 5 6 a
3 9 8 c
It’s a really common to observe similar patterns of data in our regular analysis. However, it’s often problematic to directly work with categorical behavior with machine learning algorithms. For this purpose, it’s useful to vectorize or convert these categorical behavior in some sort of indicator behavior. For example, the first row may be represented as:
num_1 num_2 a b c
0 1 2 1 0 0
Today, I learned that there exists a pandas function to do exactly this
>>> pandas.get_dummies(df)
num_1 num_2 cat_a cat_b cat_c
0 1 2 1 0 0
1 4 5 0 1 0
2 5 6 1 0 0
3 9 8 0 0 1