This initial version includes bindings for Python, Perl, PHP, and Java for use of built-in functions, models, and algorithms. OSCATS can be used from Matlab via the Java bindings. Full bindings for Matlab, as well as support for creating new models and algorithms in bound languages is targeted for version 1.0.
Features:
- IRT Models included: 1PL (Rasch), 2PL, 3PL, Nominal response
- Classification Models included: DINA, NIDA
- Selection Algorithms included:
- Random
- Optimize an arbitrary criterion
- Closest difficulty
- Maximize Fisher information
- Maximize KL information (for IRT or classification)
- Can simulate based on either IRT or classification
- Support for examinee covariates in IRT models
- Support for tagging item characteristics (e.g. "algebra," "geometry," "negative stem," etc.)
