Column selection

Select columns with bracket notation. One column returns a Series; a list of columns returns a DataFrame. Use df.filter or df.loc for advanced patterns.

Basic selection

import pandas as pd
df = pd.DataFrame({'a': [1,2], 'b': [3,4], 'c': [5,6]})
s = df['a']          # Series
sub = df[['a', 'c']] # DataFrame
print(type(s), type(sub))

Adding and renaming

• df['new_col'] = values — add or overwrite column
• df.rename(columns={'old': 'new'}) — rename (returns copy by default)
• df.drop(columns=['x']) — remove columns
• df.assign(total=lambda d: d['a'] + d['b']) — functional style

Pitfalls

Chained assignment (df[mask]['col'] = val) can fail silently—prefer df.loc[mask, 'col'] = val. Dot notation (df.col) only works for valid Python identifiers.

Important interview questions and answers

1. Q: df['a'] vs df[['a']]?
   A: Single brackets → Series; double brackets → DataFrame with one column.
2. Q: Chained assignment?
   A: Setting on a slice may not modify original—use loc with row and column together.

Self-check

1. Select two columns into a new DataFrame.
2. How do you add a computed column?

Pitfall: Chained assignment df[mask]['col']=x triggers SettingWithCopyWarning—use loc.