Advanced Pandas quiz

Question 1

What is the difference between .loc and .iloc in pandas?

Accepted Answer

.loc uses label-based indexing with inclusive end slices

Answer

.loc uses integer positions; .iloc uses labels

Answer

.loc slices are exclusive at the end; .iloc slices are inclusive

Answer

.loc and .iloc are identical when the index is a RangeIndex

Question 2

What is chained indexing and why is it dangerous?

Accepted Answer

Using .loc for assignment avoids the SettingWithCopyWarning

Answer

Chained indexing is always safe because pandas detects it

Answer

df.loc[mask, 'col'] = value may silently fail

Answer

df['col'][mask] = value always modifies the original DataFrame

Question 3

What is the difference between groupby transform and agg?

Accepted Answer

transform returns a same-length result aligned with the original index

Answer

agg returns a same-shape DataFrame with group values broadcast back

Answer

transform reduces each group to a scalar

Answer

agg and transform always return the same number of rows

Question 4

What memory-saving dtypes should you use when loading large DataFrames?

Accepted Answer

category dtype encodes low-cardinality strings as integers internally

Answer

category dtype increases memory for low-cardinality columns

Answer

float64 is always more memory-efficient than float32

Answer

pandas always infers optimal dtypes from CSV files

Question 5

How do you efficiently apply a custom function across a DataFrame?

Accepted Answer

Vectorized NumPy operations are fastest; iterrows() is slowest

Answer

iterrows() is the fastest way to process DataFrame rows

Answer

apply(axis=1) is always faster than vectorized operations

Answer

pipe() is faster than apply() for all custom transformations

Question 6

What is pd.cut() vs pd.qcut() and when do you use each?

Accepted Answer

pd.cut() creates equal-width bins; pd.qcut() creates equal-population bins

Answer

pd.cut() creates equal-population bins

Answer

pd.qcut() creates equal-width bins

Answer

Both pd.cut() and pd.qcut() always create exactly 10 bins

Question 7

How does Pandas handle time zones in datetime columns?

Accepted Answer

tz_localize attaches a timezone; tz_convert converts between timezones

Answer

tz_localize converts between timezones

Answer

tz_convert attaches a timezone to a naive datetime

Answer

Naive and aware datetimes can be mixed freely in pandas

Question 8

How does groupby with observed=True affect categorical columns?

Accepted Answer

observed=True restricts groups to categories present in the data

Answer

observed=True includes empty category groups

Answer

observed=False only shows categories present in the data

Answer

observed parameter has no effect on non-categorical columns

Question 9

What is the difference between inplace=True and reassignment in pandas?

Accepted Answer

Reassignment is preferred; inplace=True is being deprecated in spirit

Answer

inplace=True always avoids creating a copy

Answer

inplace=True enables method chaining

Answer

inplace=True is preferred by the pandas team over reassignment

Question 10

What is the pandas NA (pd.NA) and how does it differ from np.nan?

Accepted Answer

np.nan == np.nan returns False (IEEE 754 behavior)

Answer

np.nan can be used in integer columns without type promotion

Answer

pd.NA == pd.NA returns True

Answer

pd.NA is identical to np.nan in all contexts

Question 11

What are the implications of using object dtype for string columns?

Accepted Answer

object dtype stores Python string objects as pointers with high overhead

Answer

object dtype is the most memory-efficient string representation

Answer

StringDtype stores strings as Python objects like object dtype

Answer

Category dtype always uses more memory than object dtype for strings

Question 12

How does Pandas 2.0 change integer NA handling?

Accepted Answer

Nullable integer types (Int64) in pandas support missing values without float promotion

Answer

In pandas 2.0, integer columns with NaN are still promoted to float64

Answer

Nullable integer dtype is denoted with lowercase i (int64Dtype)

Answer

pd.NA behaves identically to np.nan in all operations

Question 13

What is the pandas BooleanDtype and how does it differ from bool dtype?

Accepted Answer

BooleanDtype supports pd.NA as a distinct missing value with three-valued logic

Answer

BooleanDtype converts pd.NA to False like standard bool dtype

Answer

Standard numpy bool dtype supports nullable missing values natively

Answer

BooleanDtype is slower than numpy bool for all operations

Test yourself