9 benchmark educational datasets across categories such as knowledge tracing and student performance prediction. The library provides a standardized interface for downloading and preprocessing raw student interaction logs into structured formats.
Use Cases
- Train knowledge tracing models using the standardized `user_id` and `item_id` interaction columns.
- Benchmark student performance prediction algorithms using the `score` or `correct` labels.
- Perform cross-dataset validation using the unified preprocessing interface to ensure consistent data formatting.
Strengths
- Includes 4 versions of the Assistments dataset (2009, 2012, 2015, 2017) for longitudinal study.
- Provides a standardized `user_id`, `item_id`, and `score` format across all included datasets.
- Features automated downloading and extraction scripts for large-scale educational repositories like Junyi Academy.