AI Specialization Program

🔷 WEEK 1  Lesson 2

Title: Math for Machine Learning — Vectors & Matrices (Linear Algebra Intuition)

Lesson Objective.

By the end of this lesson, learners will:

Understand why linear algebra is foundational in ML

Understand what vectors represent in real datasets

Understand datasets as matrices

Build intuition for matrix multiplication

Understand how models compute predictions mathematically

No probability.

No distributions.

Only linear algebra intuition for engineers.

1. Why Math Matters in AI Engineering

Machine Learning models are not magic.

They are mathematical functions.

When you train a model, you are:

Adjusting numerical weights

Multiplying inputs

Combining values

Producing outputs

If you don’t understand the math behind it, you cannot:

Debug model behavior

Interpret outputs

Understand scaling issues

Optimize systems

AI Engineers do not memorize formulas.

They understand structure.

Linear algebra is the structure.

2. What Is a Vector? (Engineering View)

A vector is simply:

An ordered list of numbers.

In ML, a vector represents features of one data point.

Example:

A student performance record:

[Hours Studied, Attendance %, Assignment Score]

That list is a vector.

Another example:

A house:

[Size, Bedrooms, Location Score, Age]

Every row in your dataset is a vector.

In machine learning:

Input = feature vector

Output = prediction

Vectors are how models “see” the world.

3. Features as Vectors

Every problem must be converted into numbers.

Images → pixel values

Text → embeddings (numerical representations)

Audio → frequency values

Financial data → numeric indicators

No matter the domain:

Everything becomes a vector.

This is the first mindset shift:

AI systems do not understand meaning. They understand numbers arranged in vectors.

4. What Is a Matrix?

A matrix is a collection of vectors arranged in rows.

If one vector represents one data example,

Then a matrix represents the entire dataset.

Structure:

Rows → Examples

Columns → Features

Example dataset:

| Size | Bedrooms | Location | Age |

|------|----------|----------|-----|

| 1200  |     3        |      8       | 10    |

|  900  |     2         |     6       |  15   |

This table is a matrix.

In Python (conceptually), your dataset becomes:

X = matrix of shape (number_of_examples, number_of_features)

Understanding this structure is critical.

Dataset Matrix X (3 examples, 4 features):

|  1200 |   3    |    8 |   10 | ← House 1 vector

[ 900 2 6 15] ← House 2

[1500 4 9 5] ← House 3

5. Model Weights as Vectors

A simple ML model works like this:

Prediction = Input × Weights

If input is:

[Size, Bedrooms, Location, Age]

Then the model also has:

[Weight₁, Weight₂, Weight₃, Weight₄]

The model multiplies each feature by its weight, then adds them together.

That operation is linear algebra.

6.  Matrix Multiplication Intuition

When we scale this to thousands of examples:

Instead of:

One vector × one weight vector

We do:

Dataset matrix × weight vector

This produces:

A vector of predictions.

Matrix multiplication allows models to compute:

Many predictions at once.

This is how neural networks operate internally.

Layer by layer:

Input matrix × weight matrix → output matrix.

Everything is structured multiplication.

7. How Models Compute Predictions

Let’s simplify it.

Step 1: Convert real-world problem into numeric features.

Step 2: Organize features into a matrix.

Step 3: Initialize weights.

Step 4: Multiply input matrix by weight matrix.

Step 5: Produce predictions.

That’s it.

Even deep neural networks are repeated matrix multiplications with transformations.

Linear algebra powers:

Linear regression

Logistic regression

Neural networks

Transformers

Computer vision systems

Everything starts here.

Mini Practical Exercise

Create a small dataset with:

5 rows

3 features per row

Represent:

Each row as a vector

The full dataset as a matrix

Imagine a weight vector of 3 numbers.

Manually compute:

One prediction (dot product).

Write your explanation of what happened mathematically.

Explain in 2–3 sentences what happens if one feature (e.g., 'Location Score') has much larger numbers than others—why scaling matters before multiplication.

Week 1 – Lesson 2 Outcome

Students now:

✔ Understand vectors as feature containers

✔ Understand datasets as matrices

✔ Understand matrix multiplication intuition

✔ Understand how models compute predictions

✔ Think structurally about ML systems

Next:

🔷 Week 1 – Lesson 3

Math for ML — Probability, Distributions & Debugging Intuition

And here, we introduce uncertainty properly—why predictions aren't certain, and how that affects debugging in real African datasets (e.g., noisy crop yields)."

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