Calculates IRT 2PL and 3PL models https://people.ksp.sk/~ivan/hirt
|Latest on Hackage:||0.0.1.1|
This package is not currently in any snapshots. If you're interested in using it, we recommend adding it to Stackage Nightly. Doing so will make builds more reliable, and allow stackage.org to host generated Haddocks.
Program for fitting Item Response Theory (IRT) two (2PL) and three (3PL) parameter logistic models.
Implements Joint Maximum Likelihood (JML) algorithm (currently only supported in 2PL model) and via generic function optimization using L-BFGS-B (both 2PL and 3PL).
Calculates item parameter and subject ability estimates and log likelihood statistics. For contestant abilities supports error estimates via Fisher information, and via two algorithms of the author, namely bootstrap and Bayes a posteriori probability.
Supports outputting coordinate list for a plot of Bayes a posteriori probability of individual subject abilities.
Part of a masters thesis of the author "http://people.ksp.sk/~ivan/irt/ebook.pdf" .
As a side note, it is currently a lacking proper documentation and user friendliness.
- added simple demo
- renamed statistic TaskCount to Count
- fixed nondescript error message on empty input
- relaxed version bounds on dependencies
- initial version