7. Working with Columns

df['bonus'] = df['salary'] 0.1 # create

df['bonus'] = df['salary'] * 0.1          # create
df['salary'] = df['salary'] + 1000        # update
df.drop(columns=['bonus'])                # delete (returns copy)
df.insert(1, 'rank', [1, 2, 3])           # insert at position 1

# assign: chainable new columns
df = df.assign(net=df.salary - df.tax,
               tier=lambda d: np.where(d.salary > 50000, 'high', 'low'))

df['name'] = df['name'].apply(str.upper)  # apply a function per element
df['age2'] = df['age'].map(lambda x: x*2) # map (Series only)
df[['a','b']] = df[['a','b']].applymap(float)  # elementwise on DataFrame
df.pipe(lambda d: d[d.age > 25])          # pipe for clean chaining

Method	Scope	Use
`apply`	Series or DataFrame	apply function along axis / per element
`map`	Series only	element-wise map or dict lookup
`applymap`	DataFrame only	element-wise on every cell (now `.map` on DF)
`assign`	DataFrame	add columns in a chain
`pipe`	DataFrame	insert a custom function into a chain

💡 Tip: map with a dict is a fast lookup: df['code'].map({'A':1,'B':2}).

⚠️ Common Mistake: apply with a Python function is slow. Prefer vectorized ops (df.a + df.b) or np.where when possible.

⭐ Interview Question: apply vs map vs applymap? map = element-wise on a Series; applymap = element-wise on a whole DataFrame; apply = along an axis (row/column) or per element.