Stop 01 · Programming & CS FoundationsTopics · 3/12

Functions

In short

Copy-pasting the same ten lines into five notebook cells is how bugs multiply. A function lets you write that logic once, name it, test it and reuse it everywhere.

Reusable blocks of logic with parameters, default values, *args/**kwargs and return values — how you'll structure every data pipeline and model utility.

The purple blob drops coconut and chillies into a mixer grinder next to a bowl of green chutney.

Think of it like this

The mixer grinder

Drop in coconut, chillies and salt, press the button, get chutney. You don't care how the blades spin. Change the inputs to tomato and garlic and the same machine gives a different chutney. Ingredients are arguments, the chutney is the return value.

Watch it happen

What happens when clean_name() is called

push(main script)

main script

Top of the stack is coloured

1/5Your script starts running at the top level.

The big ideas

What you need to know

Inputs in, output out

Parameters go in the brackets, return sends a value back. A function without return gives back None.

Defaults keep calls short

def split(data, test_size=0.2) lets callers skip test_size unless they want something different.

*args and **kwargs

*args collects extra positional values into a tuple. **kwargs collects named ones into a dict. Libraries use them to pass options along.

Local stays local

Variables made inside a function disappear when it returns. That keeps functions from messing with each other.

See it in code

Walk through the code

python

1def normalise(scores, low=0, high=100):2    """Squeeze scores into the 0-1 range."""3    span = high - low4    return [(s - low) / span for s in scores]5 6def describe(*values, **options):7    rounding = options.get("digits", 2)8    avg = sum(values) / len(values)9    return round(avg, rounding)10 11print(normalise([50, 75, 100]))          # [0.5, 0.75, 1.0]12print(normalise([5, 10], high=10))       # [0.5, 1.0]13print(describe(3, 4, 8, digits=1))       # 5.0

1/4low and high have defaults. The docstring explains the function in one line.

normalise is a mini version of what scikit-learn's MinMaxScaler does. describe shows how *args and **kwargs accept flexible inputs.

Save yourself the pain

Common mistakes

Oops

Using a list as a default, like def add(item, bucket=[]).

Do this instead

Use bucket=None and create the list inside. A default list is shared across every call and keeps growing.

Oops

Printing the result inside the function instead of returning it.

Do this instead

Return the value. Then the caller can print it, save it or pass it to the next step.

Your mission

Build a tiny stats toolkit

~20 min

Quick check

A function has no return statement. What does calling it give you?

Go deeper

1Practice writing functions with type hints and docstrings.
2Learn the difference between positional, keyword, default and variadic arguments.
3Refactor a long script into small, testable functions.