PISA Process Data for Complex Problem-Solving Analysis

Ruoxin Yuan's research framework, published on figshare in May 2026, includes a 22.7 MB repository of code and documentation. The work proposes a model-agnostic method for analyzing computer-based assessment log files, validated using real data from multiple PISA items. The repository is open for public use under a CC-BY-4.0 license.

Use Cases

• Extracting intrinsic features from process data to generate probability curves reflecting student progression.
• Identifying key actions within problem-solving sequences based on the proposed analytical framework.
• Uncovering response patterns in complex problem-solving tasks using deep learning models like GRU and Transformer.
• Applying a generalized and scalable analytical solution to diverse CPS items from assessments like PISA.

Strengths

• The framework is validated using real data from multiple PISA items, suggesting empirical grounding.
• Includes a user-friendly code repository (22.7 MB) open for public use, promoting reproducibility.
• The proposed probability curves are described as robust and model-agnostic statistics.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale modeling.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset files are PDF and ZIP formats, indicating the primary content may be documentation and code rather than a ready-to-use data table.

Provenance

Source

figshare, authored by Ruoxin Yuan.

Collection Method

Likely derived from log files of computer-based problem-solving assessments, specifically PISA items.

Time Range

null

Freshness

Last updated 2026-05-01 14:49:49; freshness should be verified.

Geography

null