Mining and analysis of multiple association rules between the Xining loess collapsibility and physical parameters

Area of Science:

• Geotechnical Engineering
• Data Mining Applications

Background:

• Collapsibility assessment in loess areas is costly and labor-intensive.
• Predictive models are needed to streamline loess collapsibility determination.

Purpose of the Study:

• To establish association rules between loess physical parameters and collapsibility using data mining.
• To develop simplified evaluation criteria for loess collapsibility in Xining.

Main Methods:

• Grey Relational Analysis to identify key influencing factors from 13 potential parameters.
• Apriori algorithm to discover association rules between collapsibility coefficients (δs, δzs) and identified factors.
• Analysis of 1039 loess samples from Xining.

Main Results:

• Key factors influencing loess collapsibility were identified.
• Significant association rules were found between physical parameters and collapsibility.
• Proposed evaluation criteria simplify collapsibility assessment.

Conclusions:

• Data mining effectively predicts loess collapsibility, reducing experimental costs.
• The developed criteria offer a practical approach for engineering projects in loess regions.
• Recommendations enhance construction safety in collapsible loess areas.