计算数据框中分类特征的级别

Question

I am trying to count how many levels for each categorical features in a data frame. Here is an example:

df_cat = pd.DataFrame([['green','M',10.1,'class1'],['red','L',13.5,'class2'],['blue','XL',15.3,'class1'],['red', 'M', 9, 'class1']], columns=['A','B','C','D'])

The desired output:
A 3
B 3
D 2

Answer 1

Filter columns using select_dtypes and call DataFrame.nunique :

df.select_dtypes([object]).nunique()

A    3
B    3
D    2
dtype: int64

If they're categorical columns and not objects, then this stricter filtering step would be preferred:

# Categorical column conversion.
df = df.astype(dict.fromkeys('AB', 'category'))

df.dtypes    
A    category
B    category
C     float64
D      object
dtype: object

df.select_dtypes([pd.Categorical]).nunique()
A    3
B    3
dtype: int64