Guess from Examples

Build a tiny classifier that guesses a label by finding the nearest example.

🎯 What you will learn

A classifier gives a label to new data
Nearest-example guessing uses distance
More and better examples make better guesses

👋 Let’s get started!

A classifier labels new things using old examples. The simplest method: find the example closest to the new number and copy its label. This is the idea behind many real AI systems, just on a much smaller scale. 🎯

🪜 Step by step

1. 📏 Distance

abs() gives the distance between two numbers, ignoring the sign.

print(abs(3 - 8))

2. 🔎 Find the closest

Loop through the examples. Keep the label of the closest one.

best_gap = 100
best_label = None

3. 🏷️ Guess

Put the loop in a function, then guess a new number.

print(guess(3))

🧪 Try it!

Change the code and press ▶ Run to see what happens. You can’t break anything!

Nearest example wins

examples = [(1, "cat"), (2, "cat"), (8, "dog"), (9, "dog")]
def guess(value):
    best_gap = 100
    best_label = None
    for number, label in examples:
        gap = abs(number - value)
        if gap < best_gap:
            best_gap = gap
            best_label = label
    return best_label
print(guess(3))
print(guess(7))

🎨 Unplugged activity

🎯 Make your own small example list: 6 numbers, each with a label like “low” or “high”. Test three numbers by hand before you run the code. Does the code agree with you?

Fix the guesser

best_gap = 100
for number, label in examples:
    gap = abs(number - value)
    if gap < best_gap:
        best_gap = gap
        best_label = label

📖 Key words

Tap a card to flip it over!

Classifier: Gives a label to new data 🏷️
abs(): Distance without the sign 📏
Nearest neighbour: Copying the closest example’s answer 🎯

🏆 Challenge

Change the examples so 5 is labelled “medium”, then guess 5. Press ▶ Run.

Challenge

examples = [(1, "small"), (9, "big")]

Knowledge Check Quiz

What does abs(-4) give?
How does the classifier decide?
What helps a classifier most?