numeric

skrub.selectors.numeric()

Select columns that have a numeric data type.

This selects float and integer columns but not Boolean columns.

Examples

>>> from skrub import selectors as s
>>> import pandas as pd
>>> import numpy as np
>>> df = pd.DataFrame(
...     dict(
...         f64=[1.1],
...         F64=pd.Series([2.3]).convert_dtypes(),
...         i64=[2],
...         I64=pd.Series([2]).convert_dtypes(),
...         i8=np.int8(3),
...         bool_=[True],
...         Bool_=pd.Series([True]).convert_dtypes(),
...         str_=["hello"],
...     )
... )

>>> df
   f64  F64  i64  I64  i8  bool_  Bool_   str_
0  1.1  2.3    2    2   3   True   True  hello
>>> df.dtypes
f64      float64
F64      Float64
i64        int64
I64        Int64
i8          int8
bool_       bool
Bool_    boolean
str_      object
dtype: object

>>> s.select(df, s.numeric())
   f64  F64  i64  I64  i8
0  1.1  2.3    2    2   3

Use s.boolean() to also select Boolean columns:

>>> s.select(df, s.numeric() | s.boolean())
   f64  F64  i64  I64  i8  bool_  Bool_
0  1.1  2.3    2    2   3   True   True