How does groupby work in pandas?
SHORT ANSWER
groupby splits the rows into groups by key, applies a function such as sum or count to each group, and combines the results into a new object. Named aggregation lets you compute several summaries in one call.
Explain it simply
Imagine a pile of UPI receipts. You sort them into piles by city, add up each pile, and write the totals on a sheet. That is groupby: split into piles, apply a calculation to each, combine into a table.
Example
python
import pandas as pd
df = pd.DataFrame({
"city": ["Pune", "Pune", "Delhi", "Delhi", "Delhi"],
"amount": [100, 300, 200, 400, 600],
})
print(df.groupby("city")["amount"].sum().to_dict()) # {'Delhi': 1200, 'Pune': 400}
out = df.groupby("city").agg(total=("amount", "sum"), orders=("amount", "count"))
print(out)
# total orders
# city
# Delhi 1200 3
# Pune 400 2Notes:
- Group keys become the index and are sorted by default. Use
as_index=Falseor.reset_index()to get a normal column back. agg(name=(column, function))is the cleanest way to label results.size()counts all rows per group, whilecount()ignores missing values.
What to say in the interview
"groupby follows split-apply-combine. I choose the key columns, then aggregate with sum, mean, count or several functions through agg. I use reset_index to flatten the result and know that count skips NaN while size does not."
Common mistakes
- Forgetting that the grouping column becomes the index.
- Using count when you wanted the number of rows including missing values.