Understanding Feature Importances in Football Analytics

What Are Feature Importances?

Feature importances in machine learning indicate how much each variable contributes to the model’s predictions. In football analytics, we can use them to understand which player actions or team metrics influence a player's market value the most.

Key Insights from the RandomForest Model

In my study, one of the most important features was avg_totalPullBack_value_per_game, which measures a player’s ability to successfully pull back the ball during an attack (=the 45 pass). Players excelling in this aspect tend to have a higher market value.

Another crucial metric was avg_attemptsConcededIbox_Per90_per_game, indicating how often a team concedes attempts in the penalty box. A lower number suggests strong defensive positioning, positively influencing a player's value.

Why This Matters

By identifying key factors that impact player valuation, clubs can better understand which player attributes drive transfer fees and tailor scouting strategies accordingly.

📊 Want to see which skills impact market value the most? Stay tuned for our deep dive into Partial Dependence Plots (PDPs) in the next post!