Multi-Class Classification

What Does "Multi-Class" Mean?

In Binary Classification, the model chooses between exactly two classes.

In Multi-Class Classification, there are more than two possible classes.

The model must choose one class from these possible categories.

Simple Example — Animal Classification

Imagine that we want to identify an animal from an image.

The model may have these possible classes:

Suppose we provide a new image to the model.

How Does Multi-Class Classification Work?

The model learns from training examples where the correct class is already known.

Unlike Binary Classification, the model has more than two possible answers.

Example — Fruit Classification

Suppose we want a model to identify different types of fruit.

Suppose the model receives information about a new fruit such as its weight, color, size, and shape.

The model chooses one class from the available fruit categories.

Binary vs Multi-Class Classification

The easiest way to understand Multi-Class Classification is to compare it with Binary Classification.

Example — Handwritten Digit Recognition

Another useful example is recognizing handwritten numbers.

Suppose a model needs to identify digits from 0 through 9.

There are ten possible classes.

The Model Chooses One Class

When a new input arrives, the model considers the possible classes and produces a prediction.

The important idea is that the output is one of the available categories.

Simple Python Example

Here is a small example using scikit-learn. The model predicts one of three possible classes.

from sklearn.linear_model import LogisticRegression

X = [
    [1],
    [2],
    [3],
    [6],
    [7],
    [8],
    [11],
    [12],
    [13]
]

y = [
    "Small",
    "Small",
    "Small",
    "Medium",
    "Medium",
    "Medium",
    "Large",
    "Large",
    "Large"
]

model = LogisticRegression()

model.fit(X, y)

prediction = model.predict([[7]])

print(prediction)

In this example there are three possible classes:

When the model receives the new value 7, it predicts one of those classes.

Check Your Understanding