Model Training Settings for ESG Accounting Experiments

Peng Hou's figshare release contains model training settings for a proposed feature extraction framework. The framework was tested on three real-world ESG-related accounting datasets, achieving a 4.7% average AUC improvement. The data is provided in a 5.5 KB XLS file.

Use Cases

• Analyze model configuration parameters used in experiments on three ESG accounting datasets.
• Replicate the reinforcement learning-guided steganographic encoding framework using the documented settings.
• Benchmark classification accuracy and feature redundancy reduction (12.5%) against the provided experimental parameters.

Strengths

• Experimental results are validated on three real-world ESG-related accounting datasets.
• The proposed method achieved a documented 4.7% average AUC improvement over baseline models.
• Dataset is openly shared under a CC BY 4.0 license.

Limitations

• The 5.5 KB file size suggests a very limited scope, likely containing only configuration parameters, not the underlying training data.
• No sample data or column definitions are provided, limiting understanding of the data structure.

Provenance

Source

figshare, author Peng Hou.

Collection Method

Parameters from experimental research on a novel feature extraction framework.

Time Range

null

Freshness

Last updated March 2026.

Geography

null