Aggregation & reshaping — quiz

Which keeps the original row count?

1. agg
2. transform
3. sum
4. count

Answer: transform — transform aligns the result back onto every row, like a window function.

Why call reset_index() after groupby?

1. To sort
2. To turn the grouping key back into a normal column
3. To remove nulls
4. To save memory

Answer: To turn the grouping key back into a normal column — Otherwise it stays in the index and complicates plotting and merging.

df['x'].rank() on tied values gives, by default:

1. Sequential integers
2. The average rank, e.g. 2.5
3. The lowest rank
4. An error

Answer: The average rank, e.g. 2.5 — method="min" or "dense" match SQL behaviour.

The efficient way to get the top 10 rows by revenue:

1. sort_values then head
2. nlargest(10, "revenue")
3. rank then filter
4. groupby

Answer: nlargest(10, "revenue") — nlargest avoids sorting the whole frame.

A left merge increases the row count. This means:

1. Normal behaviour
2. The right table has duplicate keys
3. Missing values
4. Wrong how=

Answer: The right table has duplicate keys — Duplicate keys on the right fan the rows out.

Which argument raises an error if the relationship is not as expected?

1. indicator
2. validate
3. suffixes
4. how

Answer: validate — validate='many_to_one' checks the key relationship.

Which converts month columns into rows?

1. pivot_table
2. melt
3. stack
4. concat

Answer: melt — melt unpivots wide data into long form.

pivot_table differs from pivot because it:

1. Is faster
2. Aggregates duplicate entries rather than raising
3. Only takes one column
4. Requires a date index

Answer: Aggregates duplicate entries rather than raising — pivot raises when the index/column pair is not unique.