Skip to main content
Colin Jaffe/2 min read

Continuous Classification: Linear Regression Explained

Regression vs Classification

FeatureWhat it predicts
Linear RegressionA continuous numeric value — price, temperature, demand.
Logistic RegressionA categorical label — yes/no, churn/stay, spam/ham — despite the name.
Master Machine Learning at Noble Desktop

Noble Desktop's Python Machine Learning Bootcamp covers scikit-learn, Keras, neural networks, and applied ML.

This lesson is a preview from our Data Science & AI Certificate Online (includes software) and Python Certification Online (includes software & exam). Enroll in a course for detailed lessons, live instructor support, and project-based training.

Use linear regression to predict continuous numeric values by fitting a line to data. Watch this tutorial to learn the key concepts and techniques.

We're going to use a linear regression model, and I want to introduce exactly why we're doing that. We are assigning a label or value to data. But there are two different kinds of classification, discrete classes and continuous classes.

Now, discrete classes are where there's a small number of possible classes. If you're like, is this a dog or a cat? Is this a mammal, reptile, fish, bird, or amphibian? Should this car have four, six, or eight cylinders? Was this car sold or unsold? There are only a couple of possible options. On the other hand, continuous classification means there's a lot of possible answers, and that's what we're looking at here.

Things like prices, or distances, or weights. These are numeric values that could be anything. It could cost $39,000.

It could cost $39,001. It could cost $39,002. And then we could dive into cents, right? It gets as granular as you want, and there aren't a finite number of numbers.

So, unlike, you know, hey, is this a dog? Is this a cat? We are asking it to predict any possible value. So that makes it a continuous classification, not a discrete classification. And that's what linear regression is for.

Linear regression, again, is going to plot a line, just like we did with attendance and concessions, and try to make predictions based on that. Okay, and we're going to eventually find this, or possibly more slopes, like we did with polynomial regression. It's going to find an equation that best predicts this value. That's what a linear regression is.

It's a line, and that's why we're doing a linear regression. Now, a funny thing with linear regression is that it's really easy to create and then train the model. There's not much code involved.

There's a lot of work. There's a lot of concepts to understand, but it's not a lot of actual code. So, let's start with instantiating a linear regression model, meaning creating it.

And the way we do that is we just say model equals linear regression, calling the function, and that's straight from scikit-learn. If we run that, great, we have a model. Now, we haven't trained it on our data yet.

That'll come next.