pandas crosstab in Python: Counts, Totals and Sums

A pandas crosstab counts how often each pair of values shows up: call pd.crosstab(rows, columns) with two columns and you get a frequency table, one row per unique value of the first and one column per unique value of the second.

pd.crosstab(df["region"], df["category"])

Every table below came from a real run on Python 3.12.5, pandas 3.0.6 and Matplotlib 3.11.2. The data is a small orders file: 12 Q1 2026 orders across three US regions, with amounts in USD.

In this tutorial, we’ll build count tables, add totals, sum and average dollar amounts, nest several columns, and check when crosstab beats pivot_table or groupby.

The sample orders data for these crosstab examples

Save this as orders.csv next to your script. If you’d rather build it in code, see creating a pandas DataFrame from a dict.

order_date,region,category,segment,amount
2026-01-05,East,Technology,Consumer,1249.99
2026-01-09,West,Furniture,Corporate,689.50
2026-01-14,East,Office Supplies,Consumer,45.20
2026-01-21,South,Technology,Corporate,899.00
2026-02-02,West,Technology,Consumer,329.99
2026-02-11,South,Furniture,Consumer,415.75
2026-02-17,East,Furniture,Corporate,1120.00
2026-02-25,West,Office Supplies,Corporate,88.40
2026-03-04,South,Technology,Consumer,189.99
2026-03-12,East,Technology,Corporate,2199.00
2026-03-19,West,Office Supplies,Corporate,64.75
2026-03-27,South,Technology,Consumer,159.99

Count combinations with a basic pd.crosstab

Pass the row labels first and the column labels second. Each cell is the number of orders with that region and category:

import pandas as pd

df = pd.read_csv("orders.csv")
print(pd.crosstab(df["region"], df["category"]))

Output:

category  Furniture  Office Supplies  Technology
region                                          
East              1                1           2
South             1                0           3
West              1                2           1

It’s the two-column version of counting unique values in a pandas column, and it saves you a groupby plus an unstack.

Add row and column totals with margins

margins=True adds a total row and column. I rename it with margins_name, and rownames/colnames label the axes:

import pandas as pd

df = pd.read_csv("orders.csv")
ct = pd.crosstab(df["region"], df["category"],
                 margins=True, margins_name="Total",
                 rownames=["Region"], colnames=["Category"])
print(ct)

Output:

Category  Furniture  Office Supplies  Technology  Total
Region                                                 
East              1                1           2      4
South             1                0           3      4
West              1                2           1      4
Total             3                3           6     12

To check that grand total against the raw data, counting rows in a pandas DataFrame gives the same 12.

Sum or average a column with values and aggfunc

By default crosstab counts. Pass a numeric column as values plus an aggfunc such as "sum" or "mean", and you’ll get dollars instead of counts. You can’t use one without the other:

import pandas as pd

pd.set_option("display.precision", 2)   # show USD with 2 decimals
df = pd.read_csv("orders.csv")
sales = pd.crosstab(df["region"], df["category"], values=df["amount"],
                    aggfunc="sum", margins=True, margins_name="Total")
print(sales)
print()
print(pd.crosstab(df["region"], df["segment"], values=df["amount"], aggfunc="mean"))

Output:

category  Furniture  Office Supplies  Technology    Total
region                                                   
East        1120.00            45.20     3448.99  4614.19
South        415.75              NaN     1248.98  1664.73
West         689.50           153.15      329.99  1172.64
Total       2225.25           198.35     5027.96  7451.56

segment  Consumer  Corporate
region                      
East       647.60    1659.50
South      255.24     899.00
West       329.99     280.88

Any combination with no rows comes back as NaN, not 0; replacing NaN with 0 in pandas fixes that. For several functions at once, see pandas groupby aggregation functions.

Build a crosstab with multiple index columns

Pass a list of columns to get a MultiIndex, meaning nested row labels. Here rows are region plus segment, and the columns are the order month pulled from the date:

import pandas as pd

df = pd.read_csv("orders.csv", parse_dates=["order_date"])
month = df["order_date"].dt.strftime("%m-%b")
print(pd.crosstab([df["region"], df["segment"]], month))

Output:

order_date        01-Jan  02-Feb  03-Mar
region segment                          
East   Consumer        2       0       0
       Corporate       0       1       1
South  Consumer        0       1       2
       Corporate       1       0       0
West   Consumer        0       1       0
       Corporate       1       1       1

Lists work for columns too. Reading dates straight from the file is covered in reading a CSV file with pandas.

Show crosstab percentages with normalize

normalize="index" turns each row into shares that add up to 1; "columns" and True work the same way down columns or over the whole table:

import pandas as pd

df = pd.read_csv("orders.csv")
print(pd.crosstab(df["region"], df["category"], normalize="index").round(2))

Output:

category  Furniture  Office Supplies  Technology
region                                          
East           0.25             0.25        0.50
South          0.25             0.00        0.75
West           0.25             0.50        0.25

Each region’s row adds up to 1.0. I go deeper, with margins and formatting, in percentage normalization using crosstab in pandas.

When should you use crosstab instead of pivot_table or groupby?

All three give the same sums here. I’ve checked it, and both comparisons print True:

import pandas as pd

df = pd.read_csv("orders.csv")
ct = pd.crosstab(df["region"], df["category"], values=df["amount"], aggfunc="sum")
pt = df.pivot_table(index="region", columns="category", values="amount", aggfunc="sum")
gb = df.groupby(["region", "category"])["amount"].sum().unstack()

print(ct.equals(pt), ct.equals(gb))

Output:

True True

Reach for crosstab for counts and for arrays or Series that aren’t in one DataFrame. Use a pandas pivot table when the data’s already a DataFrame, and groupby in pandas for long, tidy results.

Plot a pandas crosstab as a stacked bar chart

A crosstab is already a DataFrame, so .plot() draws one bar per row and stacks the columns:

import pandas as pd
import matplotlib.pyplot as plt

df = pd.read_csv("orders.csv")
ct = pd.crosstab(df["region"], df["category"], values=df["amount"], aggfunc="sum")

ax = ct.plot(kind="bar", stacked=True, figsize=(8, 4.5), rot=0)
ax.set_ylabel("Sales (USD)")
ax.set_title("Q1 2026 sales by region and category")
plt.tight_layout()
plt.show()

More chart types, including heatmaps, are in creating plots using pandas crosstab.

Next, you could try one of these:

• Stacked bar plots in pandas
• Sort a pandas DataFrame by multiple columns
• Get unique values in a pandas DataFrame
• Count duplicates in a pandas DataFrame

Frequently asked questions