Multi-Objective Sensitivity Analysis
Also known as: MOSA, Multi-criteria sensitivity analysis, Pareto sensitivity analysis, Multi-objective SA
Multi-Objective Sensitivity Analysis (MOSA) examines how changes in model parameters, weights, or assumptions affect an entire set of competing objectives simultaneously. Rather than asking how a single output shifts, MOSA tracks changes in the Pareto front or trade-off surface, revealing which parameters most destabilize multi-objective solutions and where decision-maker choices are robust versus fragile.
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 MOSA when a multi-objective optimization or simulation model contains uncertain or estimated parameters and decision-makers need confidence that the Pareto-optimal trade-offs are stable. It is especially appropriate in engineering design, health technology assessment, environmental policy, and supply chain planning where model inputs carry uncertainty and multiple conflicting goals must be balanced. Do not apply MOSA when only a single objective is relevant (use standard sensitivity analysis instead), when the model is computationally prohibitive to re-solve many times (consider surrogate-based approaches), or when parameters are known with high certainty and exploratory analysis is unnecessary.
Strengths & limitations
- Reveals which uncertain parameters most affect the structure of multi-objective trade-offs, prioritizing where to invest in better data collection.
- Identifies robustly non-dominated solutions — alternatives that remain Pareto-optimal across a wide range of parameter values — supporting more confident decision-making.
- Applicable to any multi-objective optimization or simulation framework regardless of the underlying solver or objective functions.
- Provides a richer picture than single-objective sensitivity analysis by capturing how the entire trade-off frontier shifts, not just a single metric.
- Supports transparent communication to stakeholders by showing the range of possible Pareto fronts under different assumptions.
- Computationally expensive: re-solving a multi-objective problem many times can be impractical for complex models without surrogate methods or parallel computing.
- Aggregating sensitivity across a Pareto front into a single metric (e.g., hypervolume difference) involves choices that can influence conclusions.
- Interactions between parameters are difficult to fully characterize; exhaustive factorial designs grow exponentially with the number of uncertain inputs.
- Results depend on the parameter ranges chosen for perturbation — poorly chosen ranges yield misleading sensitivity rankings.
- Does not prescribe a final decision; it only characterizes robustness, leaving the choice among trade-offs to the decision-maker.
Frequently asked
How is MOSA different from single-objective sensitivity analysis?
Single-objective sensitivity analysis tracks changes in one output metric as parameters vary. MOSA tracks changes in an entire set of trade-off solutions (the Pareto front), capturing how multiple competing objectives shift together and whether the structure of trade-offs changes — not just the magnitude of a single value.
What sensitivity metric should I use to compare Pareto fronts?
Hypervolume difference is the most common global metric; it measures the volume of objective space dominated by one front but not another. Generational distance and inverted generational distance are also used. The choice depends on whether you care about convergence, spread, or both, and whether the fronts have the same cardinality.
Can MOSA be combined with Monte Carlo methods?
Yes, and this is a common practical approach. Random parameter samples are drawn from their uncertainty distributions, the multi-objective problem is solved for each sample, and the resulting family of Pareto fronts is analyzed statistically. This yields confidence intervals on Pareto front location and shape.
When is MOSA computationally feasible?
MOSA is feasible when the underlying multi-objective solver is fast (analytical or small-scale problems) or when surrogate models (metamodels) can approximate the objective functions. For expensive simulations, adaptive sampling strategies or Gaussian process surrogates are typically required to limit re-evaluations.
Does MOSA guarantee finding a robust optimal solution?
No. MOSA diagnoses which regions of the Pareto front are sensitive to parameter uncertainty; robust multi-objective optimization methods (e.g., min-max formulations) are used to explicitly search for solutions that perform well across uncertainty scenarios. MOSA informs that step but does not replace it.
Sources
- Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D., Saisana, M., Tarantola, S. (2008). Global Sensitivity Analysis: The Primer. Wiley, Chichester. ISBN: 9780470059975
- Ehrgott, M. (2005). Multicriteria Optimization (2nd ed.). Springer, Berlin. DOI: 10.1007/3-540-27659-9 ↗
How to cite this page
ScholarGate. (2026, June 3). Multi-Objective Sensitivity Analysis. ScholarGate. https://scholargate.app/en/simulation/multi-objective-sensitivity-analysis
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.
- MONTE-CARLO-SIMULATIONDecision-making↔ compare
- Multi-objective goal programmingSimulation↔ compare
- Multi-Objective OptimizationSimulation↔ compare
- SENSITIVITY-ANALYSISDecision-making↔ compare