AI Specialization Program

WEEK  3  Lesson 1

Title: Supervised vs Unsupervised Learning

Learning Objective

By the end of this lesson, learners will:

Understand what Machine Learning is

Distinguish supervised learning from unsupervised learning

Understand the role of labels in ML

Recognize common ML problem types

Identify real-world applications of each approach

Develop the ability to frame ML problems correctly

This lesson is about Machine Learning task understanding.

Before building models, engineers must understand what type of problem they are solving.

1. What is Machine Learning?

Machine Learning (ML) is a method where computers learn patterns from data.

Instead of writing rules manually, we allow algorithms to discover patterns automatically.

Traditional programming:

Rules + Data → Output

Machine Learning:

Data + Output → Learned Rules.

The system studies examples and learns how to produce predictions.

Examples:

• Predicting house prices

• Detecting fraud transactions

• Identifying spam emails

• Recognizing images

Machine Learning is essentially pattern recognition at scale.

2. Supervised Learning

Supervised learning is the most common type of Machine Learning.

In supervised learning:

The dataset includes labels.

Labels represent the correct answers.

Example dataset:

House Size     Rooms     Price

     1200             3       200000

     1500             4       250000

Here:

Features = House Size, Rooms

Label = Price

The model learns how inputs relate to outputs.

Example problems:

• Predicting loan approval

• Detecting fraudulent transactions

• Email spam detection

• Predicting customer churn

The algorithm learns by comparing predictions to the known answers.

3. Unsupervised Learning

Unsupervised learning works without labels.

The model receives data but no correct answers.

Its goal is to discover hidden structure in the data.

Example dataset:

 Age      Income     Spending

  25       40000        300

  45       90000        700

The algorithm might discover:

• Customer groups

• Hidden patterns

• Similar behavior clusters

Common tasks:

Customer segmentation

Anomaly detection

Market basket analysis

Recommendation systems

The model organizes data based on similarity.

4. Key Difference: Labels vs No Labels

Supervised Learning:

Data + Labels → Learn Prediction

Unsupervised Learning:

Data Only → Discover Patterns

Supervised learning answers:

“What will happen?”

Unsupervised learning answers:

“What structure exists?”

Understanding this difference is the first step of ML problem framing.

5. Real-World Examples

Supervised learning examples:

Credit risk prediction

Medical diagnosis systems

Sales forecasting

Crop yield prediction

Unsupervised learning examples:

Customer segmentation

Fraud anomaly detection

Topic discovery in documents

Social network clustering

Different problems require different approaches.

6. Choosing the Right Approach

AI engineers always ask:

Do we have labeled data?

If yes → Supervised learning.

If no → Unsupervised learning.

Sometimes labels exist but are expensive to obtain.

In those cases engineers may:

Use unsupervised learning first

Then label clusters

Then train supervised models later

This hybrid strategy is common in real-world AI systems.

7.  Problem Framing Mindset

Before coding, engineers define:

What is the goal?

What data do we have?

Are labels available?

What prediction are we making?

Example:

Goal: Detect fraud.

Data: Transaction history.

Labels available? Yes.

Conclusion: Supervised classification problem.

Correct framing prevents wasted effort.

Mini Practical Exercise

Students should:

Find three real-world problems.

Classify each as:

Supervised Learning

Unsupervised Learning

Explain the reasoning.

Example:

Spam Detection → Supervised

Customer Segmentation → Unsupervised

Write a short explanation for each.

Lesson Outcome

Students now:

✔ Understand Machine Learning fundamentals

✔ Distinguish supervised vs unsupervised learning

✔ Understand the importance of labels

✔ Recognize real-world ML applications

✔ Learn how to frame ML problems correctly

This is the foundation of Machine Learning thinking.

Next:

🔷 Week 3 Lesson 2 — Regression Models

Where students build their first predictive models.

Powered by Soft AI Africa | Training the Next Generation of AI Leaders in Africa.