GroupBy Basics

What is GroupBy?

GroupBy is like sorting papers into piles and then counting each pile.

Imagine you have sales data. GroupBy lets you answer:

How much did each store sell?
What's the average sale per product?
How many orders per customer?

Basic GroupBy Example

code.py

import pandas as pd

# Sales data
sales = pd.DataFrame({
    'Store': ['NYC', 'NYC', 'LA', 'LA', 'LA'],
    'Product': ['Apple', 'Banana', 'Apple', 'Banana', 'Apple'],
    'Sales': [100, 150, 120, 180, 90]
})

print(sales)

Output:

  Store Product  Sales
0   NYC   Apple    100
1   NYC  Banana    150
2    LA   Apple    120
3    LA  Banana    180
4    LA   Apple     90

Now let's group by Store:

code.py

# Group by store and get total sales
store_sales = sales.groupby('Store')['Sales'].sum()
print(store_sales)

Output:

Store
LA     390
NYC    250
Name: Sales, dtype: int64

LA sold 390, NYC sold 250.

How GroupBy Works

GroupBy has 3 steps:

Split - Divide data into groups
Apply - Do something to each group (sum, mean, count)
Combine - Put results back together

Original Data → Split by Store → Calculate Sum → Result

NYC: 100, 150    → Sum = 250
LA: 120, 180, 90 → Sum = 390

Common GroupBy Operations

Sum - Total for Each Group

code.py

total_sales = sales.groupby('Store')['Sales'].sum()
print(total_sales)

Mean - Average for Each Group

code.py

avg_sales = sales.groupby('Store')['Sales'].mean()
print(avg_sales)

Output:

Store
LA     130.0
NYC    125.0
Name: Sales, dtype: float64

Count - How Many in Each Group

code.py

count_sales = sales.groupby('Store')['Sales'].count()
print(count_sales)

Output:

Store
LA     3
NYC    2
Name: Sales, dtype: int64

Min and Max

code.py

min_sales = sales.groupby('Store')['Sales'].min()
max_sales = sales.groupby('Store')['Sales'].max()

print("Minimum:")
print(min_sales)
print("\nMaximum:")
print(max_sales)

GroupBy on Multiple Columns

code.py

# Group by both Store and Product
grouped = sales.groupby(['Store', 'Product'])['Sales'].sum()
print(grouped)

Output:

Store  Product
LA     Apple      210
       Banana     180
NYC    Apple      100
       Banana     150
Name: Sales, dtype: int64

LA sold 210 worth of Apples.

Get Result as DataFrame

By default, groupby returns a Series. To get a DataFrame:

code.py

# Method 1: reset_index()
result = sales.groupby('Store')['Sales'].sum().reset_index()
print(result)

Output:

  Store  Sales
0    LA    390
1   NYC    250

code.py

# Method 2: as_index=False
result = sales.groupby('Store', as_index=False)['Sales'].sum()
print(result)

Same output.

Practice Example

code.py

import pandas as pd

# Employee data
employees = pd.DataFrame({
    'Department': ['Sales', 'Sales', 'IT', 'IT', 'HR'],
    'Name': ['John', 'Sarah', 'Mike', 'Emma', 'Tom'],
    'Salary': [50000, 55000, 70000, 65000, 45000]
})

print("Employee Data:")
print(employees)

# Total salary by department
print("\nTotal salary by department:")
print(employees.groupby('Department')['Salary'].sum())

# Average salary by department
print("\nAverage salary by department:")
print(employees.groupby('Department')['Salary'].mean())

# Number of employees per department
print("\nEmployees per department:")
print(employees.groupby('Department')['Name'].count())

Key Points

df.groupby('column') groups data by that column
Add ['column'].sum() or .mean() to calculate
Common operations: sum(), mean(), count(), min(), max()
Use reset_index() to get DataFrame result
Group by multiple columns: groupby(['col1', 'col2'])

Common Mistakes

Mistake 1: Forgetting to select a column for calculation

code.py

# This groups but doesn't calculate anything useful
sales.groupby('Store')

# Add what you want to calculate
sales.groupby('Store')['Sales'].sum()

Mistake 2: Using wrong column for calculation

code.py

# Can't sum text columns
sales.groupby('Store')['Product'].sum()  # Wrong!

# Sum numeric columns
sales.groupby('Store')['Sales'].sum()    # Correct

What's Next?

You learned basic groupby. Next, you'll learn about multiple aggregations - getting sum, mean, and count all at once.