How to check if a column ends with either a or b in pandas

Question

I need to filter columns by the last character, testing against multiple characters.

import numpy as np
import pandas as pd

df = pd.read_table("F:\\bridges.txt", names = ['IDENTIF','RIVER', 'LOCATION', 'ERECTED', 'PURPOSE', 'LENGTH', 'LANES', 
 'CLEAR-G', 'T-OR-D', 'MATERIAL', 'SPAN', 'REL-L', 'TYPE']) 

print(df.columns[df.columns.str.endswith('N' or 'H' or 's') ])

Output:

Index(['LOCATION', 'SPAN'], dtype='object')

Here I am not getting all columns ending with either N , H or s .

Answer 1

[col for col in df.columns if col[-1] in ['N', 'H', 'S']]

If I remember correctly, the columns attribute of a dataframe is not a series so you can't treat it as such. It's a list.

To clarify, the columns aren't technically lists. They are some variation of a special type of pandas Index. But for 99% of all intents and purposes they can be treated as lists. The point I'm trying to make clear is that they are not Series and thus don't have series methods.

Answer 2

You can use pd.Index.str.endswith with a tuple , followed by Boolean indexing:

L = ['IDENTIF','RIVER', 'LOCATION', 'ERECTED', 'PURPOSE', 'LENGTH',
     'LANES', 'CLEAR-G', 'T-OR-D', 'MATERIAL', 'SPAN', 'REL-L', 'TYPE']

df = pd.DataFrame(columns=L)

cols = df.columns[df.columns.str.endswith(tuple('HNS'))]

Index(['LOCATION', 'LENGTH', 'LANES', 'SPAN'], dtype='object')

The functionality mimics Python's built-in str.endswith , which allows you to supply a tuple to match against multiple items as alternative conditions.

Answer 3

df_serial = df_copy.filter(regex = '(?:H|N|S)$' , axis=1)
print(df_serial)

Using regular expression we can do that