Iris uses a modified Bayes Network algorithm to predict assessment responses. She updates certainty, question difficulty and skill ratings as she collects feedback. Using a text mapping technique called term frequency–inverse document frequency (TF-IDF), Iris recommends content based on your Skill IQ.

Iris builds on Item Response Theory (IRT) and applies Bayesian approximation. Learners are scored against a representative global estimate of all users of a given technology or skill, and questions are usable far sooner than with traditional IRT-based methods while maintaining similar levels of accuracy and reliability.

Iris applies Bayesian statistics to assign scores to learners and characterize questions quickly and accurately. Using a modified Glicko scoring algorithm, Iris powers adaptive assessments; the algorithm applies Markov Chain Monte Carlo methods to evaluate each learner’s skill level, providing adaptively selected questions until a high level of certainty is attained.