Hirpa SciML Academy

Play Fruit Detective with points. An APPLE is usually: Red (+10), Round (+5), Has a stem (+3) An ORANGE is usually: Orange (+10), Round (+7), Bumpy skin (+4) A BANANA is usually: Yellow (+10), Long (+8), Curved (+3) Look at a mystery fruit. Give it points for each feature. Add up the score. Highest score wins. That is how the first image classifiers worked.

Loading Python + NumPy...

Run your code to see output here.

🍎🍊🍌 Fruit Classifier

Adjust the feature weights. The computer scores each fruit and picks the winner!

Feature Weights (points)

Apple Features

Redness10

Roundness5

Has Stem3

Orange Features

Orangeness10

Bumpiness4

Banana Features

Yellowness10

Length8

Mystery AApple (124 pts)

🍎 124🍊 25🍌 0

Features: red=9, round=5, stem=3

Mystery BOrange (141 pts)

🍎 35🍊 141🍌 0

Features: round=7, orange=9, bumpy=4

Mystery CBanana (154 pts)

🍎 0🍊 0🍌 154

Features: yellow=9, long=8

Mystery DApple (86 pts)

🍎 86🍊 64🍌 26

Features: red=5, round=6, stem=2, orange=3, bumpy=1, yellow=1, long=2

This is a LINEAR CLASSIFIER. Modern neural networks do the same thing — just with millions of features and weights learned automatically from data. You just built the core idea!

You built a classifier. It adds weighted feature scores: the same idea as a neural network, with simpler math. Later tracks learn how computers find the best weights with gradient descent.