simulation_for_reddit

simulate.bad.test.for.reddit = function(j){
    accuracy = c()
    head_injuries = c(rep(TRUE, 12), rep(FALSE, 17))
    for(i in 1:174){
        guesses = sample(c(TRUE, FALSE), size=12+17, replace=TRUE)
        accuracy = c(mean(guesses==head_injuries), accuracy)
    }
    return(max(accuracy))
}

x = sapply(1:10000, simulate.bad.test.for.reddit)
hist(x, breaks=50, col='cornflowerblue', 
     main='Best (0 actual predictive power) Metabolite "Accuracy"')
print(quantile(x))

# simulation gives:
#        0%       25%       50%       75%      100% 
# 0.6551724 0.7241379 0.7586207 0.7586207 0.8965517