Small Classes

A class describes how to make a kind of object.

Here is a small class:

Create an object by calling the class:

__init__ initializes the object

__init__ runs when a new object is created. It receives the values passed to the class call:

ScoreRecord("Ada", 8)

Inside __init__, self means the new object being built.

Those lines store attributes on the object.

Keep classes small at first

A class is useful when the data has a stable shape and a name in the problem.

Good early examples:

• ScoreRecord
• ExperimentConfig
• TrainingResult

Poor early examples:

• a class that only hides one number;
• a class created because code “should use objects”;
• a class with many unrelated responsibilities.

Why not always use dictionaries?

A dictionary is excellent for flexible data. A class is useful when the fields are known and repeated.

This is easier to mistype:

record["scroe"]

With an object or dataclass, the intended field names are more visible in the definition.