Summary Dataset of Gameplay Logs, Learning Outcomes, and Adaptive Decisions from a Q-Learning-Based Early Numeracy Game for First-Grade Students

This dataset provides a summary-level release of learning outcomes, gameplay log summaries, adaptive decision summaries, reward patterns, Q-value update summaries, rule-consistency validation, and observation summaries from a Q-Learning-based adaptive early numeracy game for first-grade elementary students. The dataset was collected from a classroom field trial involving 38 first-grade students who played a six-level mathematics educational game. The game was designed to support early numeracy learning, including counting objects, number recognition, number ordering, and simple place-value understanding. The dataset includes anonymized student-level learning outcomes, gameplay event summaries, level performance summaries, student-state distributions, adaptive action distributions, difficulty transition summaries, reward summaries, Q-value update summaries, rule-consistency validation, and teacher/researcher observation summaries. This release can support learning analytics, educational data mining, adaptive educational game research, adaptive difficulty modeling, and preliminary analysis of Q-Learning-based adaptive decision mechanisms. No direct student identifiers are included. This version is a summary-level dataset; full granular event-level logs are not included in this release.