Linear Regression: Putting things in line

the_undefined_architect — Tue, 14 Apr 2026 14:27:24 +0000

Like every developer knows, Hello World is famously the first application you build when you start learning how to code. In the same spirit, house price prediction is one of the best beginner examples for understanding how machine learning models are trained and used.

Say you want to build an app that predicts house prices. How would you do it? There is no simple if-else statement that can accurately predict the price of a house in your neighborhood.

So let’s simplify the problem.

Imagine you collected data for 10 houses and wanted to explore whether house size can help us predict house price. Your dataset might look like this:

If we plot these points on an X and Y axis, this is what we get:

As we can see, a correlation emerges between the size of the house and its price (duh).
But how can we define this relationship so that, given a real value of X (the house size), we can predict the price?

This is where Linear Regression comes in.

Instead of trying to match every point perfectly, linear regression finds a line, called the regression line, that best represents the overall trend in the data.

Here’s what that looks like:

Now that we have our regression line, we can actually use it to make predictions.

Let’s say we want to estimate the price of a house that is 250 m².

We simply take that value (X = 250), project it onto our regression line, and get the predicted price:

And that’s it — we’ve trained our first model for predicting house prices.

I know, I know… we didn’t go into how the model actually finds this line, or how to implement it in code. We’ll get there in the next post.

For now, the goal was to give you an intuition for how Machine Learning works:

how models learn from data
how training shapes their behavior
and most importantly — that there is no certainty, only probability and prediction

ML Unchained: Machine Learning for Developers

the_undefined_architect — Sun, 12 Apr 2026 19:01:53 +0000

If you’re an experienced developer, you’ve probably felt it already.

Machine Learning is everywhere.
Recommendations, pricing, search, fraud detection, and copilots.

And yet — it still feels… separate from what you do.

Like a different world.

Why You Should Care

Not because it’s hype.
Because it changes how you build.

As an experienced developer, you’re used to writing deterministic systems.

Given X → return Y

You structure it with:

Functions
Classes
Tests

Machine Learning doesn’t replace that.
But it introduces a component that doesn’t follow explicit rules — it predicts.

Instead of writing logic, you train it.
Instead of exact outputs, you get probabilities.
Instead of debugging code, you analyze behavior.

Why This Transition Is So Hard

Traditional developers struggle to get started with ML because the AI world feels intimidating. There’s no simple “hello world” to ease you in.

It’s not that simple — but it’s not that hard either.

I’ll say it clearly: you don’t need to be a math wizard to become an ML engineer. I’m not. Yes, math is involved, but as a developer you’ve already dealt with harder problems.

The real issue is how people approach learning it.

They think they need to:

Learn the math first
Then the theory
And only then start coding

That’s backwards.

We’re engineers.

We don’t learn by reading first —
we learn by building cools shit.

Final Thoughts

That’s exactly what we’re going to do together.

We’re going to dive into Machine Learning — but not through theory-first learning.

We’ll go project first.

We’ll build things.
We’ll break them.
We’ll understand how they behave.

And along the way, the concepts will start to make sense — naturally.

No unnecessary complexity.
No waiting until you’re “ready.”

We'll do it the engineering way, and we'll do that 500 words at a time

Forem: the_undefined_architect

Linear Regression: Putting things in line

ML Unchained: Machine Learning for Developers

Why You Should Care

Why This Transition Is So Hard

Final Thoughts