Inspiration

I am a passionate Business Intelligence Analyst and Football Fan. Every time a team decides to sign, or not to keep, a player, there is a wide range of reactions and criticism. It is often difficult to differentiate between team mismanagement and strategic management. I aimed to build a model I have never seen before in my years a a fan- a machine learning model that analyses the best salary cap distribution needed to maximize regular season wins.

What it does

Analyses relationship between percentage of salary cap allocation among starters and regular season wins.

How I built it

  1. Proprietary data set. I obtained salary information for every team for the past five years, calculated their salary cap percentage of team's total cap, and flagged starters.
  2. Create and analyze Machine learning models in R (decision tree, random forest, and multiple linear regression).
  3. Create dashboard to see and interactive with results.

Challenges I ran into

  1. Obtaining data. The data set I needed was not publicly available. Only part that was available was the salary information, but I need to calculate their salary as a percentage of the total and flag starters based on my personal football knowledge.
  2. Models were off at first. Made a small error creating initial data set.
  3. Analyzing Machine Learning models is inherently time consuming

Accomplishments that I'm proud of

  1. Have something to present!
  2. Learned something about football I did not previously know.
  3. Created a brand new NFL related data set I can share with others.

What I learned

  1. Something I did not realize until I had done this exercise is that EXTERIOR offensive linemen are crucial to the success of a team whereas INTERIOR offensive linemen are surprisingly the least important position.
  2. I learned how to integrate multiple machine learning models into a single R dashboard.

What's next for Optimal NFL Salary Distribution

  1. Continue to polish. My goal is to be able to produce a product that is useful to someone who knows nothing about statistics.
  2. Apply my models to teams after the free agency ends and the NFL draft takes place. I will be using them throughout next season to help me out in fantasy football!

Built With

Share this project:
×

Updates