Ranking Features to Promote Diversity: An Approach Based on Sparse Distance Correlation

Andi Wang, Juan Du, Xi Zhang, Jianjun Shi

Research output: Contribution to journalArticlepeer-review


The improvement of sensing technology enables features of process variables to be collected during the fabrication of products. This article develops an automatic tool for process feature rankings based on these data. Based on the sensing data characteristics and the need of manufacturing system analysis, we propose two rules of the feature ranking scheme: assessing general dependency between each individual process feature and the quality variable, and satisfying a diversity rule. Specifically, we propose a feature ranking scheme based on the sparse distance correlation (SpaDC) that satisfies these two rules. Theoretical properties of the proposed algorithm are investigated. Simulation studies and two real-case studies from semiconductor manufacturing applications demonstrate that the SpaDC method ranks the features effectively given these two ranking rules.

Original languageEnglish (US)
StateAccepted/In press - 2022


  • Distance correlation
  • Feature ranking
  • Manufacturing process
  • Sparsity

ASJC Scopus subject areas

  • Statistics and Probability
  • Modeling and Simulation
  • Applied Mathematics


Dive into the research topics of 'Ranking Features to Promote Diversity: An Approach Based on Sparse Distance Correlation'. Together they form a unique fingerprint.

Cite this