VGFTS-AI Proof-of-Concept Simulation: Code and Data

Simulation code and generated data underlying the proof-of-concept study reported in Section 5 of the article "VGFTS-AI: An Adaptive Intelligence Layer for Personalized Virtual Reality Serious-Game Training Through Behavioral Learner Modeling." The study reports no human-subject data; the human evaluation described in the article is a proposed, pre-registerable protocol that has not yet been run. This deposit contains a fully reproducible (fixed random seed) Python simulation of the proposed adaptive loop, in which synthetic learners drive a random-forest proficiency estimator and a Learner Digital Twin, together with the generated data files (learning curves, twin-tracking trajectories, and per-learner outcomes for the full, estimator-only, and static conditions) underlying the paper's Figures 4–5, Table 7, and reported metrics. See the included README for file descriptions and instructions to reproduce all results. Requires Python with numpy, scikit-learn, and matplotlib.