News By Tag
News By Location
MCHT, an R package for Monte Carlo and bootstrap hypothesis testing, is now available on GitHub
MCHT is an R package for Monte Carlo and bootstrap hypothesis testing. It is available for free under the MIT license on GitHub.
By: Curtis Miller's Personal Website
MCHT is a package for facilitating Monte Carlo and bootstrap hypothesis testing, statistical methods that rely on random number generation for deciding between statistical hypotheses rather than analytically derived mathematical functions that often are appropriate only for large samples or under strong distributional assumptions. This approach to statistical testing can produce exact or conservative tests at any sample size, or can at least have better finite sample properties than large-sample tests.
The package has a primary function, MCHTest(), that can accept R functions that specify the test and then returns a function that can conduct the test. The function is intended to be easy to use and flexible while still providing decent performance by taking advantage of parallelization on multi-core computers. The test returned functions much like R functions like t.test(), thus giving a familiar interface.
Curtis Miller has begun publishing articles demonstrating how to use the package, starting with the announcement article available at: https://ntguardian.wordpress.com/
The package is considered to be in a beta stage and thus Curtis Miller invites others to use the package and report problems that emerge. It is licensed for use under the MIT license. R users who can use the devtools package can easily install it using the command devtools::install_