Grey Clustering: Whitenization-Based Classification Under Uncertainty
Grey Clustering (Grey Incidence / Whitenization) · Also known as: Grey Whitenization Weight Function Clustering, Grey Fixed-Weight Clustering, Grey Variable-Weight Clustering, Gri Kümeleme
Grey Clustering is a classification method from grey systems theory that assigns objects to predefined grey classes using whitenization weight functions. Developed within the framework of Deng Julong's grey system theory and systematized by Sifeng Liu, it is particularly suited for situations involving small sample sizes, incomplete information, or uncertain data—conditions common in engineering assessments, environmental monitoring, and socioeconomic evaluation. The method quantifies how strongly each object belongs to each grey class and makes a crisp assignment based on maximum clustering coefficients.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Grey Clustering when you need to classify objects into ordered qualitative categories under data scarcity or uncertainty, and when criteria weights and class boundaries can be specified from domain knowledge or standards. It fits well in environmental quality assessment, project risk grading, supplier evaluation, and infrastructure condition rating. Assumptions include meaningful turning-point specifications for each class and commensurable criteria scales. It is not appropriate for exploratory clustering where the number of classes is unknown (use k-means or hierarchical clustering instead), and it requires careful elicitation of whitenization parameters to avoid arbitrary results.
Strengths & limitations
- Works well with small or incomplete datasets where probabilistic methods lack statistical power
- Whitenization functions can directly encode regulatory standards or expert-defined class boundaries
- Produces both object-level and system-level classification summaries via the comprehensive clustering index
- Handles partial, uncertain, or qualitative measurements without requiring distributional assumptions
- Whitenization turning points and criterion weights must be specified a priori, introducing subjectivity
- The number of grey classes must be predetermined; the method does not discover natural cluster structure
- Sensitive to the choice of turning points—poorly calibrated functions can yield misleading class assignments
- Not suited for high-dimensional, large-scale datasets where machine learning clustering methods are more efficient
Frequently asked
How is Grey Clustering different from Fuzzy C-Means?
Fuzzy C-Means iteratively optimizes cluster prototypes from the data itself, making it exploratory and data-driven. Grey Clustering uses predefined whitenization weight functions whose turning points reflect external standards or domain expertise. Grey Clustering is confirmatory—it asks how well objects fit known categories—whereas Fuzzy C-Means discovers latent structure. The two methods share a soft-membership philosophy but differ fundamentally in setup and purpose.
What is a whitenization weight function and how do I set its turning points?
A whitenization weight function maps a criterion value to a membership degree in a given grey class, rising from zero, plateauing at full membership, then declining back to zero across four turning points. The turning points are typically drawn from regulatory standards, expert consensus, or calibrated from historical data. Poorly chosen turning points directly undermine classification validity, so domain consultation and sensitivity testing are essential before finalizing the function parameters.
Can Grey Clustering handle mixed quantitative and qualitative criteria?
Yes. Qualitative criteria are first mapped to numerical scales (e.g., Likert or coded ordinal scales) and then treated like any quantitative criterion within the whitenization framework. The weight assigned to each criterion can reflect its relative importance. This flexibility makes Grey Clustering practical for multi-criteria problems in management and policy where not all indicators are directly measurable on continuous scales.
Sources
- Liu, S., & Lin, Y. (2010). Grey Systems: Theory and Applications. Springer. ISBN: 978-3-642-13937-6
How to cite this page
ScholarGate. (2026, June 2). Grey Clustering (Grey Incidence / Whitenization). ScholarGate. https://scholargate.app/en/soft-computing/grey-clustering
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Fuzzy C-MeansMachine learning↔ compare
- GM(1,1) Grey ForecastingSoft Computing↔ compare