Transfer Trend Analytics

One of the big projects I did with a group in the graduate program was in applied machine learning (IST 707) regarding transfers in association football (soccer). Our group sought to analyze how soccer clubs try to maximize wins by spending large fees for star players, or those that seek to make more on their bottom line by selling these top prospects to the big buyers. Our dataset came from Kaggle and related many different teams across European leagues competing in many different competitions, sometimes against one another, and other times separated entirely. Many of the larger, more lucrative clubs tend to play each other more often, and the smaller tiered teams can almost be viewed as developmental for players on track to play at a higher level. We carried out many machine learning algorithms in Python to see which factors amount to maximizing wins for clubs, or what drives larger fees for certain star players. My portion was to focus on the factors that relate to winning games on the pitch, rather than the monetary figures in play off the field. The biggest factor in maximizing wins was the age of visiting teams. Across many sports, winning away from home has proven difficult, especially with younger, inexperienced players. This wasn’t necessarily noteworthy to anyone that understands the details of sport, however, something like this superseding how many national team players a team has or how much they’ve spent on players, age matters more. Moreover, age was one of, if not the most significant factor in the prices clubs pay for these star players. The learning outcome best described for this project would be “apply visualization and predictive models to help generate actionable insight” with our models trying to predict and visualize what matters most in maximizing wins or profits for sporting organizations.