Cabral, Pedro da Costa BritoPimenta, Rodrigo Vicente2023-11-212023-11-212023-10-23http://hdl.handle.net/10362/160216Internship Report presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Business AnalyticsThis thesis focuses on the analysis of player transfers in football, with a specific emphasis on Benfica, the Portuguese club known for its significant investments in player acquisitions. The study aims to determine a market value that aligns closely with actual transfer values while identifying the key features and factors that play a crucial role in this context. By understanding these factors, the research aims to predict the market value of the players, in order to help the club's decision-making process. To achieve these objectives, the data was collected from various leagues, including the Portuguese, English, and French leagues, which are of interest to Benfica's scouting department. The TransferMarket website served as the data source, with data extraction performed using the BeautifulSoup library in Python. The analysis phase involved exploring and identifying the most relevant variables and their impact on calculating the market value using three modeling techniques: Decision Trees, Random Forest, and XGBoost. This process was conducted twice, once for the global dataset and once specifically for Benfica's purchases, allowing for a comparison between the two scenarios. The conclusions revealed that players who have previously played in the Premier League or Eredivisie and are international players for their respective national teams are highly prioritized by Benfica. Furthermore, factors such as the time remaining on a player's contract and their age were identified as important variables in determining market value. The Random Forest model exhibited the best performance, with a higher r-squared value (0,919 for Benfica dataset and 0,953 for the Global dataset) and a lower Mean Absolute Error (3818086,11 for Benfica dataset and 4038199,56 for the Global dataset) for both the TransferMarket and Actual Transfer Value problems, respectively. This study points to the significance of player transfers in professional football and highlights the potential benefits of employing data-driven approaches for decision-making by clubs. By incorporating these methodologies and leveraging comprehensive information, clubs can make more informed decisions and improve their overall strategic planning.engSports AnalyticsMarket ValueData AnalysisDecision TreesRandom ForestXGBoostDecision Makingmaster thesis203390385