A comparison of local explanation methods for high-dimensional industrial data: A simulation study

作者：

Highlights：

• Simulated high-dimensional process-like datasets with binary quality variable.

• Real-world process data with a simulated response.

• Evaluated twelve prediction and explanation approaches using two metrics.

• Generalized linear models perform well for monotone relationships.

• Tree-based models are robust for multiple types of relationships.

摘要

•Simulated high-dimensional process-like datasets with binary quality variable.•Real-world process data with a simulated response.•Evaluated twelve prediction and explanation approaches using two metrics.•Generalized linear models perform well for monotone relationships.•Tree-based models are robust for multiple types of relationships.

论文关键词：Local Explanations,Simulation,Shapley values,Interpretable model,Statistical process control

论文评审过程：Received 5 January 2022, Revised 25 May 2022, Accepted 16 June 2022, Available online 27 June 2022, Version of Record 2 July 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117918