GroupBy: Split-Apply-Combine

Reviewed & published by Brayan K

GroupBy in pandas is the operation that splits a DataFrame into groups by the values in one or more key columns, applies a summary function such as mean or sum to each group, and combines the answers into one tidy result.

Part of the free Pandas course at LearnCodingFast — hands-on lessons with examples you run in your browser, plus practice exercises and a quick quiz.

Learn the split-apply-combine pattern, group by single and multiple keys, run common aggregations, iterate groups, and reach into a specific group with get_group.

What You'll Learn in This Lesson

• The split-apply-combine mental model
• Group and summarise with mean, sum, count
• Group by multiple keys at once
• Run several functions with .agg()
• Iterate groups and use get_group()
• Keep keys as columns with as_index=False

1 Split-Apply-Combine in Action

Every groupby follows three steps. Split the rows into groups by a key, apply a function to each group, then combine the results. The pattern df.groupby("key")["value"].mean() reads exactly like that sentence.

• Split: rows are bucketed by the unique values in the key column
• Apply: a function like mean, sum, or count runs on each bucket
• Combine: one row of output per group is stitched together

import pandas as pd

sales = pd.DataFrame({
    "region": ["North", "South", "North", "South", "North"],
    "product": ["A", "A", "B", "B", "A"],
    "amount": [100, 80, 50, 90, 120],
})

# Average amount per region
print(sales.groupby("region")["amount"].mean())
# North -> 90.0, South -> 85.0

# Total and count per region
print(sales.groupby("region")["amount"].sum())    # North 270, South 170
print(sales.groupby("region")["amount"].count())  # North 3, South 2

# ✅ Expected output:
# region
# North    90.0
# South    85.0
# Name: amount, dtype: float64
# region
# North    270
# South    170
# Name: amount, dtype: int64
# region
# North    3
# South    2
# Name: amount, dtype: int64

2 Multiple Keys and .agg()

Group by several columns by passing a list, which forms a group for every unique combination and produces a MultiIndex result. The .agg() method lets you run more than one function at once, returning a column for each.

import pandas as pd

sales = pd.DataFrame({
    "region": ["North", "South", "North", "South", "North"],
    "product": ["A", "A", "B", "B", "A"],
    "amount": [100, 80, 50, 90, 120],
})

# Group by TWO keys -> a result per region+product combination
print(sales.groupby(["region", "product"])["amount"].sum())

# Several aggregations at once with .agg()
print(sales.groupby("region")["amount"].agg(["sum", "mean", "max"]))
#         sum   mean  max
# North   270   90.0  120
# South   170   85.0   90

# ✅ Expected output:
# region  product
# North   A          220
#         B           50
# South   A           80
#         B           90
# Name: amount, dtype: int64
#         sum  mean  max
# region                
# North   270  90.0  120
# South   170  85.0   90

3 Iterating Groups, get_group, and as_index

A groupby object is iterable: each loop gives you the group key and its DataFrame slice. get_group() jumps straight to one group, and as_index=False keeps the grouping keys as ordinary columns instead of moving them into the index.

import pandas as pd

sales = pd.DataFrame({
    "region": ["North", "South", "North", "South"],
    "amount": [100, 80, 50, 90],
})

grouped = sales.groupby("region")

# Iterate: key + the sub-DataFrame for that group
for key, sub in grouped:
    print(key, "has", len(sub), "rows, total", sub["amount"].sum())

# Jump to a single group
print(grouped.get_group("North"))

# as_index=False keeps 'region' as a normal column
print(sales.groupby("region", as_index=False)["amount"].sum())
#   region  amount
# 0  North     150
# 1  South     170

# ✅ Expected output:
# North has 2 rows, total 150
# South has 2 rows, total 170
#   region  amount
# 0  North     100
# 2  North      50
#   region  amount
# 0  North     150
# 1  South     170

🎯 Your Turn: Sales per Region

Replace the blank so you total the amount for each region. The expected output is North 270 and South 170.

import pandas as pd

df = pd.DataFrame({
    "region": ["North", "South", "North", "South", "North"],
    "amount": [100, 80, 50, 90, 120],
})

# 🎯 YOUR TURN: group by 'region' and total the 'amount'
totals = df.groupby("___")["amount"].sum()
print(totals)

# Expected output: North -> 270, South -> 170

📋 Quick Reference

🎯 Mini Challenge: Store Report

Build a per-store report: total sales, average sale, and number of transactions, with the store kept as a column.

import pandas as pd

df = pd.DataFrame({
    "store": ["A", "B", "A", "B", "A"],
    "category": ["food", "food", "toys", "toys", "food"],
    "sales": [40, 25, 60, 30, 55],
})

# Total, average, and count of sales for each store, store as a column
report = (df.groupby("store", as_index=False)["sales"]
            .agg(["sum", "mean", "count"]))
print(report)

# Bonus: total sales per store AND category combination
print(df.groupby(["store", "category"])["sales"].sum())

# ✅ Expected output:
#   store  sum       mean  count
# 0     A  155  51.666667      3
# 1     B   55  27.500000      2
# store  category
# A      food        95
#        toys        60
# B      food        25
#        toys        30
# Name: sales, dtype: int64

❓ Frequently Asked Questions

Lesson complete — you can summarise by category!

You now understand split-apply-combine, can group by one or many keys, run mean, sum and count, apply several functions with agg, iterate groups, reach into one with get_group, and flatten results with as_index=False.

Practice quiz

Continue this course

• Previous: Working with Dates & Times
• Next: Aggregation & Transformation — Apply multiple aggregations and group-aware transforms in one step
• Quick reference: Pandas cheat sheet › Grouping & Aggregation

Frequently asked questions

What is split-apply-combine in pandas?

Split-apply-combine is the mental model behind groupby. Pandas first splits the rows into groups based on a key column, then applies a function such as mean or sum to each group, and finally combines the per-group results into a single output Series or DataFrame.

How do I group by more than one column in pandas?

Pass a list of column names, for example df.groupby(['region', 'product']). Pandas forms a group for every unique combination of those columns and returns a result with a MultiIndex, one level per grouping key.

What does as_index=False do in groupby?

By default the grouping keys become the index of the result. Passing as_index=False keeps them as ordinary columns instead, which gives you a flat DataFrame that is easier to merge or export. It is equivalent to calling reset_index afterwards.