TUTOR SESSION: Machine Learning Foundations

CURRENT CONCEPT: Supervised Learning

EXPLANATION
Supervised learning means we have examples (inputs + correct answers) and train a model to predict outputs for new inputs. Think of learning to cook: you see recipes (input) and taste the result (correct answer). After seeing many recipes and results, you learn patterns about what ingredients work together.

KEY IDEAS
→ Training data: labeled examples (house size + sale price)
→ Model learns patterns: bigger house → higher price
→ Prediction: show new house size → model predicts price
→ Accuracy: how often predictions match reality

EXAMPLE WALKTHROUGH
Dataset: 100 houses
- Input: square footage (500-5000 sq ft)
- Output: price ($100k-$1M)
Model learns: price increases roughly $200 per sq ft

TAKE IT FURTHER
Question for you: If a house is twice as big, would the price be exactly twice as much? Why or why not?
(Hints available below)

CHECK YOUR UNDERSTANDING
1. Name two real-world problems that could use supervised learning
2. What's the difference between input features and output labels?
3. If your training data only had small houses ($100k-300k), would predictions on a $2M mansion be reliable? Why?

PRACTICE PROBLEM
You have customer spending data (age + purchase amount). Build a supervised model to predict spending for new customers. What could go wrong?

HINT: Consider outliers (rich 80-year-old, broke 25-year-old)
Difficulty: medium
Time estimate: 10 minutes

WHEN READY
Press continue → next topic (Unsupervised Learning) or ask deeper question