如何使用.loc便捷选取DataFrame的连续与非连续列？

Great question—when you're working with large DataFrames, writing concise column selection code saves a lot of time. Your current approach works, but there are more streamlined methods to avoid building the column list step-by-step.

Here are a few handy alternatives:

1. Unpack the Column Slice Directly in the List #

You can use the unpacking operator (*) to convert the column slice into individual elements, then add your standalone column(s) right in the list passed to .loc:

import pandas as pd

df = pd.DataFrame({'a': [1], 'b': [1], 'c': [1], 'd':[1], 'e':[1]})
df.loc[:, [*df.columns['a':'c'], 'e']]

This is super readable and lets you do the selection in one line—no need to create an intermediate list variable.

2. Use pd.Index.union() for Safe Combination #

If you want to ensure no duplicate columns (useful in more complex scenarios), you can combine the slice and standalone columns using union():

cols = df.columns['a':'c'].union(['e'])
df.loc[:, cols]

This method automatically removes any duplicate column names, which is a nice safeguard if your standalone columns might overlap with the range.

3. Combine with np.r_ (Using Column Positions) #

If you prefer working with column positions instead of names, you can use numpy.r_ to concatenate the position ranges, then use .iloc (a useful alternative even though you asked for .loc):

import numpy as np

col_positions = np.r_[df.columns.get_loc('a'):df.columns.get_loc('c')+1, df.columns.get_loc('e')]
df.iloc[:, col_positions]

Why Your Original Idea ['a':'c', 'e'] Doesn't Work #

Just to clarify: Python lists don't support slicing syntax inside them (like ['a':'c']), which is why that approach throws an error. The methods above work by converting the column slice into a proper iterable of column names first, then combining it with your standalone columns.

内容的提问来源于stack exchange，提问作者Edward