Understanding machine learning model evaluation metrics — Part 4

Navigating the Metrics Landscape: A Comprehensive Guide to Evaluating Classification Models in Machine Learning

Introduction

In the previous articles, we introduced various metrics to evaluate performance of a model typically used in classification problems and regression problems. You can find these here: Part 1, Part 2 and Part 3. In this article, we will focus on some more common metrics that are used to evaluate the performance and accuracy of machine learning models for regression problems:

• Mean Percentage Error (MPE)
• Mean Absolute Percentage Error (MAPE)
• R-Squared (R²)

We will see what each metric means, how to calculate it, and how to interpret it. We will also show how to implement these metrics. Let’s get started!

Mean Percentage Error (MPE)

Mean Percentage Error (MPE) is a metric that measures the average percentage of the errors between the predicted values and the actual values. It is calculated as the sum of the percentage differences between the predictions and the observations, divided by the number of samples. The formula for MPE is:

where yi​ is the actual value, y^​i​ is the predicted value, and n is the sample size.