Soccer Analytics with Machine Learning

Chapter 4. Regression Techniques for Soccer Analytics

In Chapters 2 and 3, we built the foundations of the analytical workflow used throughout this book. We set up the environment, learned the core tools, and used exploratory analysis to understand the data. The next step is to move from description to prediction.

This chapter introduces regression, the first predictive framework we will present in depth. Regression is used when the outcome we want to estimate is numerical rather than categorical. In soccer, that leads to questions such as these:

• What is a player’s market value likely to be next season?

• How many goals can we expect a team to score in its next match?

• How many kilometers will a midfielder cover in a game?

These are all questions that regression models can help us answer. Regression does not remove uncertainty, but it gives us a structured way to estimate expected values, compare influences, and understand how outcomes change as inputs change. That makes it useful for both forecasting and explanation.

By the end of this chapter, you will be able to do the following:

• Understand the core concepts of regression and how it applies to soccer.

• Build, train, and interpret a linear regression model to predict player market values.

• Use Poisson regression to model the number of goals in a match, a classic soccer analytics problem.

• Implement K-nearest neighbors (KNN) regression to find and analyze similar players.

• Evaluate your models to understand how well ...