What machine learning is
Rules vs learning
A normal program follows rules a person wrote: "if the order is over $50, give free delivery". But some questions don't have rules anyone can write down. How many customers will come on Saturday? It depends on the weather, the day, holidays, how the café is growing… and how they all combine.
Machine learning turns it around: you give the computer examples (two years of days, with how many customers came), and it learns the pattern itself. The result is a model — a program that can make predictions for new cases.
| Normal programming | Machine learning | |
|---|---|---|
| You give it | rules | examples (data) with answers |
| It gives you | answers | a model that has learned the rules |
| Good for | clear rules: tax, totals, logins | fuzzy patterns: forecasts, spam, images, language |
The main kinds
| Kind | Predicts | Corner Café example |
|---|---|---|
| Regression | a number | how many customers will come on Saturday |
| Classification | a category | will this member come back? is this review happy? which digit is this? |
| Clustering | groups nobody named | what kinds of customers are there? |
Regression and classification are supervised learning: the examples come with the right answers. Clustering is unsupervised: the computer finds groups on its own.
Where you meet it
Streaming recommendations, spam filters, bank fraud alerts, weather forecasts, photo search on your phone, predictive text — and the large language models behind AI chatbots, which are huge machine learning models trained on text. Everything in this course is the foundation those are built on.
What it can't do
- It only knows what's in its data. Bad or biased data means bad or biased predictions
- It finds patterns, not causes
- It's never perfect — the skill is knowing how wrong it is, and whether that's good enough
Try it
- List three predictions a business you know could make with machine learning. Is each regression, classification or clustering?
- Download the course files from this lesson.
Downloads
1 file to download once you enrol.