TL;DR: a logistic regression model that I built gives us an 88% chance against SFA.
@culallen GITT
@culallen GITT
Long version: I developed a basic logistic regression model for a Kaggle competition using regular season and tourney data from 2003-2017. It has a cross validation accuracy score of about 72% (not great but I don't have time to try to improve it). I used the predict_proba function to compare all the matchups and fill out my bracket. Somewhat alarmingly, 88% is the worst score for a 3-14 matchup.
IF we beat SFA, we have a 52% chance against UF and a 66% chance against St. Bonaventure. And UF only has a 62% chance against the bonnies.
IF we beat SFA, we have a 52% chance against UF and a 66% chance against St. Bonaventure. And UF only has a 62% chance against the bonnies.
A&M only has a 47% chance against Providence
