This project explored different Machine Learning (ML) techniques to predict and study the market value of professional football players based on their characteristics and football attributes and compare it with their actual transfer value, to determine whether a player is overvalued, undervalued, or accurately valued and to also check the features that affects the market value the most. I explored several model design ideas and evaluate their performances against benchmark techniques.
In this project, I built a predictive model using machine learning regression algorithms to predict the market value of professional football players. The performance of the model compares favorably to other ML algorithms model and traditional techniques.
The predictive model was also used to identify under and overvalued players relative to the market. When specifically looking at undervalued players, it was noted that fame and popularity might be an important factor missing from the model. From these conclusions, one can conclude that the prediction of players market value using machine learning methods can help market values manipulations, as large disparities between actual and predicted results were observed in the project.