In zachmayer/kaggleNCAA: Simulate the NCAA tournament

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)
library('kaggleNCAA')
data(sample_submission_men)
write.csv(sample_submission_men, 'seed_benchmark_men.csv', row.names=F)

Kaggle NCAA Bracket Simulator

Simulate the NCAA tournament based on a kaggle-format bracket (with predictions for every possible matchup).

First, re-install the package:

devtools::install_github('zachmayer/kaggleNCAA')

Now load the tournment from a csv

seed_benchmark_men.csv is a simple seed-based benchmark I made. Turn it into a bracket with 4 functions:

• parseBracket to load the data from a .csv
• simTourney to simulate the tourney (you can also use walkTourney). Use at least 1000 simulations (more is better). One day I will parallelize this =D
• extractBracket to extract a bracket from the simulation results
• printableBracket to actually print the bracket

(Load my data with data(sample_submission_men, package='kaggleNCAA') if you want to just use the seeds)

set.seed(1)
library('kaggleNCAA')
dat <- parseBracket('seed_benchmark_men.csv', w=0)  # w=0 for men
sim <- simTourney(dat, 100, progress=TRUE, w=0)  # w=0 for men
bracket <- extractBracket(sim)
printableBracket(bracket)

If simulation's not your thing (e.g. your predicted probabilities are transitive), you can also "walk" forward through the tournament, which is much faster:

bracket <- walkTourney(dat)
printableBracket(bracket)

Note in this case that the probabilities associated with the team making it to a given round will be incorrect

zachmayer/kaggleNCAA documentation built on May 4, 2019, 9:03 p.m.