Supervised & Unsupervised Learning

Learning Objectives

By the end of this lesson, you will:

• Understand supervised learning with labeled data
• Learn about classification and regression
• Understand unsupervised learning without labels
• Learn about clustering techniques
• Compare supervised vs unsupervised learning
• See real-world applications of both types

Supervised Learning

ℹ️ Definition Supervised Learning happens when machines learn from data that has labels (correct answers).

Think of it like learning with a teacher who shows you the right answers!

In machine learning, the "teacher" is the labels that guide the learning process.

Learning from Labeled Examples

Let's understand this with a fruit example:

Imagine you're taught:

• Round + Red = Apple 🍎
• Small + Round + Purple + Bunch = Grapes 🍇

Now, when you see this fruit:

You know it's an Apple because it's round and red! This is exactly how supervised learning works.

📝 Key Point In supervised learning, we provide the machine with examples AND their correct answers (labels) to help it learn patterns.

Examples of Labeled Datasets

📊 BMI Dataset (Body Mass Index)

Weight(kg)	Height(cm)	BMI	Label ✅
54	172.7	18	Underweight
68	172.7	24	Healthy
86	172.7	29	Overweight

🌸 Iris Flower Dataset

Sepal Length	Sepal Width	Petal Length	Petal Width	Label ✅
5.1	3.5	One.4	0.2	Setosa
4.9	3.0	One.4	0.2	Setosa
6.0	2.2	4.0	One.0	Versicolor

Two Types of Supervised Learning

Supervised learning has 2 main categories:

1️⃣ Classification - Sorting into Groups

2️⃣ Regression - Predicting Numbers

Let's explore each one!

📂 Classification

ℹ️ Definition Classification predicts which category or group something belongs to.

Think of it like sorting things into different boxes!

Examples of Classification:

• Gender: Male OR Female
• Exam Results: Pass OR Fail
• Email Type: Spam OR Not Spam

Example: Email Classification

Here's how email classification works:

1. 📧 Feed Examples - Show the model many spam and non-spam emails
2. 🔍 Learn Patterns - Model finds common patterns in spam emails
3. ✅ Make Predictions - New emails are classified based on learned patterns

Real-World Applications

• 🍎 Fruit Classification - Apple vs Orange vs Banana
• 🎨 Color Detection - Red vs Blue vs Green
• 📸 Face Recognition - Identify different people

Popular Classification Algorithms

• K Nearest Neighbor (KNN) - Classifies based on nearest examples
• Decision Tree - Makes decisions like a flowchart
• Naive Bayes - Uses probability to classify

📈 Regression

ℹ️ Definition Regression predicts numbers based on patterns in data.

Think of it like predicting your test score based on how many hours you study!

Examples of Regression:

• 💰 Price: RM1,500 or RM200,000
• ⚖️ Weight: 70kg or 200kg
• 🌡️ Temperature: 25°C or 35°C

Example: Predicting Humidity from Temperature

Here's the step-by-step process:

One. Collect Data 📊 We gather temperature and humidity readings:

2. Find the Pattern 🔍 The model draws a "line of best fit" through the data:

3. Make Predictions 🎯 Use the line to predict new values:

Real-World Applications

• 🏠 House Prices - Predict based on size, location, rooms
• 📏 Height Prediction - Estimate height from age and weight
• 📈 Stock Prices - Forecast future values

Popular Regression Algorithms

• Linear Regression - Finds a straight line pattern
• Multiple Linear Regression - Uses multiple factors to predict

🆚 Regression vs Classification

Feature	📈 Regression	📂 Classification
What it predicts	Numbers (continuous values)	Categories (groups)
Output type	Numerical	Categorical
Examples	Price: RM1,234 Height: 165cm	Fruit: Apple Email: Spam
When to use	When you need a number	When you need a category
Algorithms	Linear Regression Multiple Linear Regression	KNN Decision Tree Naive Bayes

💡 Remember

• Regression = Predicting numbers 📊
• Classification = Sorting into groups 📁

Unsupervised Learning

ℹ️ Definition Unsupervised Learning happens when machines learn from data without labels (no correct answers given).

Think of it like exploring a new place without a map - you have to find patterns on your own!

Without a "teacher," the machine must:

• Explore the data 🔍
• Find hidden patterns 🎯
• Group similar things 📊

The machine looks at lots of data and finds natural groupings or patterns by itself.

Examples of Unlabeled Datasets

Notice: No labels or answers! The machine must find patterns itself.

📊 BMI Dataset (No categories given)

Weight(kg)	Height(cm)	BMI
54	172.7	18
68	172.7	24
86	172.7	29

🌸 Iris Flower Dataset (No flower types given)

Sepal Length	Sepal Width	Petal Length	Petal Width
5.1	3.5	One.4	0.2
4.9	3.0	One.4	0.2
6.0	2.2	4.0	One.0

📝 Key Difference Unlike supervised learning, we don't tell the machine what the correct answers are!

🎯 Clustering

ℹ️ Definition Clustering groups similar things together without being told what the groups should be.

Think of it like organizing your toys - you might put all the cars together, all the dolls together, without anyone telling you to!

How Clustering Works:

• Finds data points that are similar
• Groups them together
• Each group has members with shared traits

Example: Fruit Clustering

Here's the process:

1. 🍎🍊🍇 Give Mixed Fruits - Model sees fruits but doesn't know their names
2. 👀 Compare Features - Looks at shape, color, size
3. 📦 Create Groups - Puts similar fruits together
4. 🆕 Sort New Fruits - Places new fruits in the most similar group

Real-World Applications of Clustering

🛍️ Market Segmentation Group customers by shopping habits:

📺 Streaming Services

• Netflix groups viewers by what they watch
• Suggests shows based on your viewing group

🏥 Healthcare

• Group patients with similar symptoms
• Find disease patterns

Popular Clustering Algorithms

• K-Means Clustering - Creates K number of groups
• Mean-Shift Clustering - Finds dense areas of data points

Summary

In this lesson, you learned:

Supervised Learning 👨‍🏫

• Uses labeled data (with answers)
• Two types: Classification (groups) and Regression (numbers)
• Like learning with a teacher

Unsupervised Learning 🔍

• Uses unlabeled data (no answers)
• Main type: Clustering (finding groups)
• Like exploring on your own

💡 Quick Comparison

• Have labels? -> Use Supervised Learning
• No labels? -> Use Unsupervised Learning
• Need groups? -> Use Classification or Clustering
• Need numbers? -> Use Regression

Learn More

📚 Explore more: Supervised vs Unsupervised Learning

Watch and Learn

📺 Supervised vs Unsupervised Learning Explained

📝 Practice with AI Try these prompts:

• "Explain supervised vs unsupervised learning with real examples"
• "When should I use classification vs regression?"
• "What are common clustering applications?"

---

🎯 Next Lesson: Machine Learning Process - Building Your First ML Model