Worked examples & research cases
See the problem.
Inspect the reasoning.
Explore how a declared method connects inputs, results and evidence. Each case identifies its origin and limits, so an archived research adaptation and a synthetic exercise can be read on their own terms.
Archived research adaptation
A hospital supplier comparison,
with an inspectable evidence record.
- The problem
- Compare three suppliers against five criteria and examine the basis for a fixed pairwise ordering.
- Inputs and source
- The archive retains 15 printed AHP-derived local priorities from Table 3 of Akcan and Güldeş (2019): logistics, quality, cost, flexibility and reliability. These values are dimensionless priorities, not raw hospital measurements.
- The declared adaptation
- The benchmark replaces the source weighting and downstream pipeline with oriented min–max normalization, recomputed CRITIC weights and the declared CECE rankers. The additive result shown here belongs to that adaptation.
- Evidence to inspect
- The bundle includes packaged comparison records and the declared scenario region. The explorer displays the archived material; it does not rerun the original certificate construction or checker.
CRITIC–Additive · Archived result
The declared ordering
| Rank | Alternative | Score |
|---|---|---|
| 1 | Supplier 1 | 0.626359 |
| 2 | Supplier 3 | 0.450362 |
| 3 | Supplier 2 | 0.357815 |
Synthetic teaching example
Make criterion importance
visible before ranking.
- The problem
- Establish criterion weights from variation and conflict in an illustrative supplier performance matrix.
- Inputs and method choice
- Six alternatives and four criteria: cost, quality, delivery and reliability. The example uses CRITIC, which combines dispersion and signed correlation after the declared benefit/cost transformation.
- What can be examined
- Trace the criterion calculations in Weighting Lab, then use explicit weight or data scenarios in Sensitivity Lab to examine changes to the resulting decision model.
- Scope and review
- The matrix and example expert judgements are illustrative. The displayed weights describe these inputs and this formulation. They are separate from the three-supplier CECE archive above; they carry no completed human review.
CRITIC · Six-alternative example
The resulting weights
| Criterion | Weight |
|---|---|
| Cost | 0.497482 |
| Quality | 0.160971 |
| Delivery | 0.162526 |
| Reliability | 0.179021 |
Synthetic worked example
A high score cannot
replace a mandatory condition.
- The problem
- Document a supplier decision while keeping mandatory constraints, preferences and open evidence gaps visible.
- Inputs and approach
- The supplied six-alternative scenario combines a performance matrix, hard feasibility bounds, weights and a structured decision-evidence register.
- The example finding
- Five alternatives meet the declared baseline bounds and one is excluded. Ranking and sensitivity are interpreted within that feasible set, alongside the assumptions and available evidence.
- Scope and review
- This is a demonstration of the documented workflow. It is not a real supplier recommendation, an independently validated universal assurance score or an expert approval.
What the case helps reveal
Keep feasibility, ranking and evidence together.
- Apply non-negotiable limits explicitly.
- Record why weights and trade-offs are appropriate.
- Test the declared uncertainty scenarios.
- Keep missing evidence and reviewer decisions visible.
Guided assessment example
An AI evaluation needs
more than a performance score.
- The problem
- Examine an AI system’s intended use while distinguishing measured performance from contextual claims and missing evidence.
- Inputs and approach
- Use the seven-component, 23-subcomponent evidence workspace, with optional binary-prediction data for applicable technical measurements.
- What the record shows
- Computed measurements, researcher-supplied assessments, evidence gaps and any separately recorded expert decision have distinct statuses. No universal trust score is assigned.
- Scope and review
- The evidence prompts are a research-informed operationalization. Security, explanation quality, fairness and organizational practices require suitable evidence for the stated setting.
A useful distinction
Measurement, interpretation and review each have a role.
A prediction file can support applicable performance measurements. A specialist review examines the supplied evidence and declared scope; it is recorded against an exact analysis revision.
Explore the specialist review processApply the approach to a real problem
Bring the evidence from your setting.
Define the decision, data rights, suitable method and review scope before interpreting the result as an application finding. Published-source adaptations and synthetic exercises provide inspectable starting points; a real application needs its own evidence.
The seven-day trial starts when the first project is created, includes one project and requires no card. There is no automatic charge after the trial. View plans and availability.