Non-linear Best Worst Method
Also known as: Non-linear BWM, Nonlinear BWM
Non-linear BWM is a variant of the Best Worst Method that replaces the linear programming formulation with non-linear optimization. Instead of minimizing the maximum deviation (Chebyshev distance), it minimizes the sum of squared deviations (L2 norm). This provides more flexible weight derivation and better accommodates uncertain or fuzzy preferences.
Key highlights
- Accommodates fuzzy or interval-valued preferences more naturally than linear BWM
- Produces smoother weight variations; no sharp jumps in weights due to small preference changes
- Utilizes all preference information; least-squares methods make full use of preference magnitudes
- Better suited to uncertain decision-making; can embed confidence or uncertainty directly in the optimization
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use Non-linear BWM when preference information is inexact or when you want to incorporate uncertainty in best-worst judgments. It is particularly suited to fuzzy decision-making, interval-based preferences, or situations where stakeholders are uncomfortable with crisp preference values. Avoid it when computational efficiency is critical or when a Chebyshev (minimax) criterion is theoretically justified.
Strengths & limitations
- Accommodates fuzzy or interval-valued preferences more naturally than linear BWM
- Produces smoother weight variations; no sharp jumps in weights due to small preference changes
- Utilizes all preference information; least-squares methods make full use of preference magnitudes
- Better suited to uncertain decision-making; can embed confidence or uncertainty directly in the optimization
- Increased computational complexity; non-linear optimization requires iterative solvers and may not guarantee global optimality
- Sensitivity to initial solutions; different starting points can lead to different local optima
- Less interpretable objective function; least-squares minimization has less transparent meaning than minimax
- Requires careful numerical implementation; ill-conditioning can arise if preferences are nearly inconsistent
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How do I express fuzzy or interval preferences in Non-linear BWM?
Represent intervals as lower and upper bounds in the objective function, or define membership functions for fuzzy preferences. The least-squares objective can directly incorporate these: minimize sum of (preference_lower-satisfaction + preference_upper-satisfaction) squared.
Which optimization algorithm should I use?
Sequential Least Squares Programming (SLSQP), trust-region methods (e.g., dogleg), or Levenberg-Marquardt are suitable. Use multiple starting points and compare solutions to ensure global optimality. Python's scipy.optimize module provides good implementations.
How much does Non-linear BWM differ from standard linear BWM?
If preferences are crisp and consistent, both methods typically yield similar weights. Non-linear BWM excels when preferences are uncertain, intervals, or fuzzy. Conduct sensitivity analysis to compare linear and non-linear results for your specific problem.
Sources
- 1.Rezaei, J. (2015). Best-worst multi-criteria decision-making method: Some properties and a linear model. Journal of Cleaner Production, 229, 976-985.
- 2.Ghorabaee, M. K., Zavadskas, E. K., Olfat, L., & Turskis, Z. (2017). Extended EDAS method for fuzzy multi-criteria decision-making: An application to supplier selection. International Journal of Computers Communications & Control, 11(3), 358-371.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Non-linear Best Worst Method. ScholarGate. https://scholargate.app/decision-making/non-linear-bwm