Welcome to AA Bracketology! This bracket is based on a XG Boost machine learning model that uses data including SOR, KPI, BPI, Kenpom, and NET ratings from previous seasons to train the model and then predict this year’s bracket. In a world where analytics is becoming more present, this bracketology method looks to see what a bracket could look like if only analytics were used to make it.
Once the teams are selected, they are placed into one of the 4 regions: East which is played in Boston this year, Midwest which is in Detroit, South which is in Dallas, or West which is in Los Angeles. Also, each team is assigned a first round location: Indianapolis, Brooklyn, Salt Lake City, Memphis, Spokane, Pittsburgh, Charlotte, or Omaha.
The data used for the model and predictions is from the toRvik package in R studio. The table is made from the gt package in R. I am also using the cbbdata package to access the team logos.
Check out this bracket and other brackets at Bracket Matrix: https://bracketproject.blogspot.com/2023/11/the-2024-bracket-matrix-links.html
Check out daily updates on my Twitter: @austin_ambler
Leave a comment