NFL 2015 Data Visualization
Professional sports have been revolutionized in recent years by the emergence of big data. Data science is a powerful tool for sports because the available information is abundant, easily categorized, and almost exclusively numeric. Data analysis has confirmed many traditional beliefs while exposing the short-sightedness of others, and revealed profound truths that have long escaped expert notice.
The National Football League (NFL) is the highest grossing American sports league, raking in $13 billion in 2016 alone. Personal fantasy football expenditures in the United States (basically legalized gambling on the NFL) are estimated to be $15 billion per year. In a market so big, data-driven insights are extremely valuable. Using a dataset with every NFL play from the 2015 season, I created a shiny app that enables users to explore individual and team performance. Using this tool, everyone from fiery coaches to couch-potato gamblers has a chance to find the extra edge they're looking for.
Aggregate Team Statistics
The first section of the app allows the user to rank team performance based on various statistical categories. One important use of this feature is to see how certain metrics are correlated with one another:
First Down Likelihood
The pursuit of first downs is fundamental to the architecture of every play. It is always advantageous for coaches (and gamblers!) to be aware of in-game probabilities of specific events, and first downs in particular. Based on the 2015 data we can estimate each team's likelihood of eventually attaining a first down in various situations:
Player Performance Biases
Constructing a football team and designing its plays is a delicate art. It is crucial that teams are aware of players' idiosyncrasies -- individual performance can vary significantly based on factors such as down, field location, and direction of the play. The following graph shows how often and how successfully Marshawn Lynch runs the ball to the left, center and right: