pythonpandasnumpydtype

Get names of all numeric columns in a pandas DataFrame (filter by dtype)

Which pandas methods can be used to get names of the columns of a given DataFrame that have numeric dtypes (of all sizes, such as uint8, and not just 64-bit ones), using a single line of code? Note: in practice there are hundreds of columns, so dtypes (or another vectorized method) should be used for detecting data types.

import numpy as np
import pandas as pd

test_df = pd.DataFrame(data=[{"str_col": "some string",
                              "int_col": 0,
                              "float_col": 3.1415}])

test_df.dtypes[test_df.dtypes == np.dtype('float')].index.values[0]
# "float_col"

test_df.dtypes[test_df.dtypes == np.dtype('int')].index.values[0]
# "int_col"

# ?
# ["float_col", "int_col"]
Solution

  • To get the numerical columns:

    numeric_cols = test_df.select_dtypes(include=['number']).columns.tolist()

    Result:

    ['int_col', 'float_col']