How to select all columns except one in pandas?

Targeted column exclusion strategies

Using difference method

To weed out specific columns, especially in dataframes larger than a Hogwarts enrollment scroll, use difference:

df = df[df.columns.difference(['Voldemort', 'Malfoy'])]  # Because who likes the villains, right?


In this example, difference plays Quidditch to bring back all columns, except the snitches.

Inverting with ~

~, the honey badger of Pandas, simply doesn't care about the columns you want to exclude:

df = df.loc[:, ~df.columns.isin(['unwanted_column'])] # Mimics Dobby's "Master has given Dobby a column"!


isin concocts a boolean cauldron with True for the to-be-excluded columns. ~ heads off to Beauxbatons, returning everyone else.

Previewing columns with drop

A quick peak into your dataframe, with a twist of exclusion:

print(df[df.columns.drop('Dolores Umbridge')]) # Because who wants Dolores in their sight?


This beauty doesn't tamper with df, it merely vanishes the 'Dolores Umbridge' column from the view.

Sticking the exclusion with inplace

To slam the door permanently on the columns:

df.drop('unwanted_column', axis=1, inplace=True) 


The original df has been Obliviated of the 'unwanted_column' - for good! Yes, just like removing a Horcrux.

Mastering important facets

Taming the SettingWithCopyWarning beast

A hair-raising side-effect of dabbling in exclusion is SettingWithCopyWarning. Often, you won't know if you took a bite off the original dataframe or merely its reflection. The antidote? .copy()

new_df = df.loc[:, ~df.columns.isin(['unwanted'])].copy()  # Now, no mimblewimble!


Avoiding "The Cursed Child"

As your Python grows up, pandas phases out some old friends, with .ix being one example. Stick to the timeless classic: .loc and its younger sibling, .iloc.

Organising column burial

Brew yourself a potion of favourite columns to dispose of, and smoothly channel your curses:

cols_to_exclude = ['temp', 'auxiliary', 'unwanted']
df = df.loc[:, ~df.columns.isin(cols_to_exclude)]


Much like a marauder's map of columns you don't want to bump into.