Insightful skiing: developing explainable models of on-snow performance through physical attribute selection of...

Area of Science:

• Sports Engineering
• Materials Science
• Biomechanics

Background:

• On-snow alpine ski evaluation is crucial but resource-intensive and subjective.
• Objective physical measurements are available but not fully leveraged for performance prediction.

Purpose of the Study:

• To develop an automated methodology for predicting on-snow ski performance using physical attributes.
• To identify a minimal set of physical properties that define ski performance.

Main Methods:

• Employed elastic net regression, bootstrap resampling, and intelligent feature selection.
• Utilized extensive physical measurements and on-snow evaluation metrics from 192 alpine skis.
• Validated the predictive models across 10 ski categories.

Main Results:

• Achieved an average Mean Absolute Error rank prediction of 15%.
• Identified key performance-defining properties using fewer than three physical attributes on average.
• Demonstrated promising predictive capabilities for ski performance.

Conclusions:

• The automated methodology offers a simple, effective, and comprehensive approach to ski performance assessment.
• Findings have significant implications for alpine ski design and consumer guidance.
• The method facilitates integration of diverse evaluation sources for continuous refinement.