Comparative Statistical Table of Prediction Models for Loess Compressibility

Zhiqi Liu's comparative statistical table evaluates established prediction models for loess compressibility. The dataset, shared on figshare in a 9.5 KB XLS file, provides error metrics and model performance comparisons. It focuses on study areas including Huocheng and Xinyuan counties in Xinjiang.

Use Cases

• Compare error metrics priority across models like Random Forest, SVM, and MLP for loess compressibility prediction.
• Analyze correlations (e.g., significantly positively correlated, negatively correlated) between mechanical parameters and model outputs.
• Evaluate model performance for engineering classification tasks using the provided statistical comparisons.

Strengths

• Dataset is openly licensed under CC BY 4.0.
• Focuses on a specific geotechnical engineering problem: loess compressibility in Xinjiang.
• Compares multiple established prediction models including Random Forest, SVM, and CART.

Limitations

• Extremely small file size (9.5 KB) suggests a very limited table, likely with few rows of summary data.
• No column names, sample data, or row count are provided, limiting understanding of the data structure.
• Tags suggest the data may be aggregated results from a study, not raw experimental measurements.

Provenance

Source

figshare, authored by Zhiqi Liu.

Collection Method

Comparative statistical table from a study evaluating prediction models.

Freshness

Last updated March 23, 2026.

Geography

Study areas include Huocheng and Xinyuan counties in Xinjiang, China.