<![CDATA[The discrepancy between market value and actual player performance often creates financial inefficiencies in football club recruitment strategies. This study aims to identify undervalued players (high performance but low valuation) using Big Data Analytics on the Transfermarkt dataset. The initial dataset is large-scale, comprising 32,601 player records and 1,706,806 match appearance entries, reflecting high-volume data characteristics The research methodology follows four systematic stages: (1) massive data acquisition and integration covering match statistics and transfer history; (2) implementation of Feature Engineering to convert raw statistics into per-90-minute metrics while accounting for contract duration; (3) fair value modeling using the CatBoost Regressor algorithm optimized with Log-Transformation to handle skewed data distributions; and (4) model validation using 5-Fold Cross Validation and residual analysis to detect price anomalies. The results demonstrate the model's ability to precisely identify potential player segments overlooked by standard market valuations. It is concluded that integrating CatBoost with robust feature engineering serves as a strategic instrument for club management to enhance investment efficiency (Return on Investment). ]]>
Copyrights © 2026
