Pandas get dummies() for numeric categorical data

Question

I have 2 columns:

• Sex (with categorical values of type string as 'male' and 'female')
• Class (with categorical values of type integer as 1 to 10)

When I execute pd.get_dummies() on the above 2 columns, only 'Sex' is getting encoded into 2 columns. But 'Class' is not converted by get_dummies function.

I want 'Class' to be converted into 10 dummy columns as well, similar to One Hot Encoding.

Is this expected behavior? Is there an workaround?

Answer 1

You can convert values to strings:

df1 = pd.get_dummies(df.astype(str))

Answer 2

If you don't want to convert your data, you can use 'columns' argument in get_dummies. Here is quick walkthrough:

Here is the data frame reproduced per your description:

sex_labels = ['male', 'female']
sex_col = [sex_labels[i%2] for i in range(10)]
class_col = [i for i in range(10)]
df = pd.DataFrame({'sex':sex_cols, 'class':class_col})
df.sex = pd.Categorical(df.sex)

The dtypes are:

print(df.dtypes)
sex      category
class       int64
dtype: object

Apply get_dummies:

df = pd.get_dummies(df, columns=['sex', 'class'])

Verify:

print(df.columns)

Output:

Index(['sex_female', 'sex_male', 'class_0',
'class_1','class_2','class_3','class_4','class_5',
'class_6','class_7','class_8','class_9'],dtype='object')

Per the docs at, https://pandas.pydata.org/pandasdocs/stable/reference/api/pandas.get_dummies.html,

If columns is None then all the columns with object or category dtype will be converted

This is the reason you only see dummies for sex column and not for class.

Hope this helps. Happy learning!

Note: Tested with pandas version '0.25.2'