Feature Importances in the XGBoost Model – What Drives Market Value?

How XGBoost Determines Player Valuation

XGBoost (Extreme Gradient Boosting) is a powerful machine learning algorithm that ranks features based on their contribution to model predictions. Unlike RandomForest, XGBoost uses a boosting technique, where weak models are refined iteratively to improve overall accuracy.

Key Feature Importances from XGBoost

📊 Most Influential Metrics:1️⃣ avg_bigChanceCreated_per_game – Creating high-quality goal-scoring opportunities is the strongest indicator of a player's market value in the XGBoost model.2️⃣ avg_attOpenPlay_value_per_game – Contribution to open play attacking sequences also plays a significant role in determining a player's worth.3️⃣ Team_points – As expected, team success remains a crucial factor, reinforcing that market values are not just based on individual stats but also team performance.

Comparing with RandomForest

XGBoost’s ranking of avg_bigChanceCreated_per_game as the top feature suggests that offensive creativity is weighted more heavily in this model compared to RandomForest, where defensive stability also played a major role.

Why This Matters for Clubs

Understanding feature importance helps clubs optimize scouting strategies. If a team is looking for undervalued talent, focusing on players with high chance creation and open play attacking influence could be a data-driven approach.

📈 Next up: A deep dive into PDPs in the XGBoost model – How do these features interact with market value?