How do I count how often each value appears in a column?

The short version

Call value_counts() on the column. It returns each distinct value with how many times it appears, sorted from most to least common. Add normalize=True for proportions instead of raw counts, and dropna=False if you want missing values counted too.

A worked example

import pandas as pd

df = pd.DataFrame({"colour": ["red", "blue", "red", "red", "blue"]})

print(df["colour"].value_counts())

colour
red     3
blue    2
Name: count, dtype: int64

Proportions instead of counts

df["colour"].value_counts(normalize=True)

colour
red     0.6
blue    0.4
Name: proportion, dtype: float64

This is the quick way to answer "what share of rows are red" without dividing by the total yourself.

Counting missing values too

By default value_counts ignores NaN. If a missing value is itself worth counting, include it.

df["colour"].value_counts(dropna=False)

value_counts versus groupby

value_counts is a shortcut for the common case of counting one column. When you need to count within groups, or count and aggregate at the same time, groupby is the general tool.

# same idea, spelled with groupby
df.groupby("colour").size()

The difference is that value_counts sorts by frequency, while groupby().size() keeps the group order. Use value_counts for a fast look, and groupby when the count is one part of a larger summary.

Counting combinations

To count pairs, pass a list of columns to value_counts on the whole frame (pandas 1.1 and later).

df.value_counts(["colour"])   # extend the list for more columns

This gives frequencies for each combination of the listed columns, which is handy for spotting how two categories co-occur.