Skip to main content
Welcome to the first major category of machine learning! Supervised learning is intuitive because it mimics how we often learn new things in real life: by studying examples with known answers.

​
The Core Idea: Learning with an Answer Key

The “supervised” in supervised learning means that the data we use to train the model is labeled. Think of it as training data that comes with a built-in “answer key.” The algorithm’s job is to learn the relationship between the input data and the correct output labels.

​
Analogy: Learning with Flashcards

Imagine you’re teaching a toddler to identify different fruits using flashcards.
  1. You show them a picture of a banana (the input data).
  2. You say the word “banana” (the label).
  3. You repeat this process with pictures of apples, oranges, and grapes, each time providing the correct label.
After showing them hundreds of these labeled flashcards, the toddler (the model) starts to learn the patterns—yellow and curved is a banana, red and round is an apple. Eventually, you can show them a picture of a new banana they’ve never seen before, and they can correctly identify it. That’s the essence of supervised learning!

​
The Two Main Tasks of Supervised Learning

Supervised learning problems can almost always be broken down into one of two categories: Classification or Regression. The difference is simply what you are trying to predict.

​
1. Classification: Predicting a Category

In a classification task, the goal is to predict a discrete label. In other words, you’re trying to put the input data into a specific category or class. The output is a label, not a number. The key question is: “What class does this belong to?” Examples:
  • Spam Detection: Is this email spam or not spam?
  • Image Recognition: Is this a picture of a cat, a dog, or a bird?
  • Medical Diagnosis: Does this patient’s scan show a malignant or benign tumor?
  • Sentiment Analysis: Is this customer review positive, negative, or neutral?

​
2. Regression: Predicting a Continuous Value

In a regression task, the goal is to predict a continuous numerical value. You’re not predicting a category, but a quantity. The key question is: “How much?” or “How many?” Examples:
  • House Price Prediction: Based on features like square footage, number of bedrooms, and location, what is the price of this house?
  • Weather Forecasting: What will the temperature be tomorrow?
  • Stock Price Prediction: What will be the price of a certain stock next week?
  • Sales Forecasting: How many units of a product will we sell next quarter?

​
Key Takeaways for Supervised Learning

  • It is task-driven; you are trying to predict a specific outcome.
  • It requires labeled data to train the model.
  • Use Classification for predicting categories (e.g., yes/no, red/green/blue).
  • Use Regression for predicting numbers (e.g., prices, temperatures, counts).