PyInns - Home

Grouping by Multiple Variables in Pandas – Multi-Level GroupBy Best Practices 2026

Grouping data by multiple variables (columns) is one of the most powerful features in Pandas. It allows you to create rich, multi-dimensional summaries such as sales by region and month, user behavior by category and country, or performance by department and quarter.

TL;DR — Modern Multi-Variable Grouping

Use groupby([col1, col2, ...]) for multiple grouping keys
Combine with date components using .dt accessor
Use named aggregation with .agg() for clean output
Apply method chaining for readability

1. Basic Grouping by Multiple Columns

import pandas as pd

df = pd.read_csv("sales_data.csv", parse_dates=["order_date"])

# Group by region and category
summary = (
    df
    .groupby(["region", "category"])
    .agg(
        total_sales=("amount", "sum"),
        avg_sale=("amount", "mean"),
        order_count=("amount", "count"),
        unique_customers=("customer_id", "nunique")
    )
    .round(2)
    .reset_index()
)

print(summary)

2. Advanced Multi-Level Grouping with Date Components

multi_summary = (
    df
    .groupby([
        "region",
        "category",
        df["order_date"].dt.to_period("M").rename("month")
    ])
    .agg(
        total_sales=("amount", "sum"),
        average_order_value=("amount", "mean"),
        transaction_count=("amount", "count"),
        unique_customers=("customer_id", "nunique")
    )
    .round(2)
    .reset_index()
)

print(multi_summary)

3. Grouping by Multiple Variables Including Time Hierarchy

report = (
    df
    .groupby([
        "region",
        df["order_date"].dt.year.rename("year"),
        df["order_date"].dt.quarter.rename("quarter")
    ])
    .agg(
        total_revenue=("amount", "sum"),
        avg_revenue=("amount", "mean"),
        num_orders=("amount", "count")
    )
    .round(2)
    .reset_index()
)

4. Best Practices in 2026

Use a **list of columns** inside groupby() for multiple grouping keys
Combine categorical columns with date components extracted via .dt
Always use named aggregation (`total_sales=("amount", "sum")`) for clear column names
Use method chaining to keep complex groupings readable
Apply .reset_index() at the end when you need a flat DataFrame
Round results for professional-looking reports

Conclusion

Grouping by multiple variables unlocks deep insights from your data. In 2026, mastering multi-level groupby() with named aggregation allows you to create sophisticated, multi-dimensional summaries with clean and maintainable code. Whether you are analyzing sales by region + product category + month or user engagement by country + device type, this technique is essential for effective data manipulation.

Next steps: