In this article, we’ll explain the concept of polynomial regression and show how it can lead to overfitting. We’ll also discuss some techniques you can use to avoid overfitting. These include using k-fold cross-validation or a hold-out set but, most importantly, we’ll discuss how applying domain knowledge will help you avoid overfitting. We won’t discuss any code by you can find the full project on GitHub.

What is Polynomial Regression?

Let’s dive straight into the concept by fitting some Linear Regression models to a dataset. We will be using a real estate valuation data set which contains information on 414 houses sold. To keep things simple, we’ll only be looking at two variables — house price of unit area and the age of the house. We can see the relationship between these two variables in Figure 1 below. The idea is to use the age of the house to try to predict the price.

#polynomial-regression #machine-learning #data-visualization #towards-data-science #data-science