Chapter 3 Remote case-based approach to supplier evaluation3.1. Approaches to Case-Based Decision Analysis Much progress has been made in case-based approaches to MCDA in recent years. As suggested in [14], the set of cases can include (1) cases on which the DM has made a decision in the past; (2) fictitious but realistic alternatives; or (3) a representative subset of the actual set of alternatives, A. One major advantage of case-based reasoning is that "decision makers may prefer to make exemplary decisions rather than explain them in terms of specific parameters of the functional model." To obtain these advantages, it is useful for the cases to be familiar enough to the DM to be easily evaluated. Additionally, the case set should contain a wide range of cases and should be neither too small nor too large. 3.1.1 Case-Based Approaches to Classification and Sorting Generally, case-based approaches to preference elicitation in MCDA include the three steps shown in Figure 3.11. Represent: Identify representative cases from the entire set of alternatives, or elsewhere, and present them to the DM for preference evaluation.2. Infer: find preference parameters that reproduce the DM's judgments about representative cases as accurately as possible.3. Evaluate: If the preference parameters are sufficient, apply them to the entire set of alternatives to obtain preferences. Then, with the appropriate tools, you can tackle the selected problem.Fig. 3.1 Case-based approach to MCDA The case-based approaches that have been proposed can be roughly grouped into two classes, depending on whether they depend on models with explicit or implicit preferences. For an explicit preference model such as the UTA method [15] or the case-based distance approach [16]...... middle of the paper ...... analyze multiple criteria by classifying problems based on the proposal Case-based distance models are shown in Figure 3.2. It includes the following steps: • Identify the set of alternatives: All possible alternatives must be considered within the appropriate limits. • Construct the criteria set: Construct a criteria set to reflect the DM's concerns and goals. • Select case set: Choose a small case set from the alternative set and ask experts to provide a ranking on this case set. • Establish the case-based distance classification model: build the classification model, optimize it, and obtain the criterion weights. • Check the feasibility of the model: use the weight information to calculate the distance of the case set and check whether the ranking order is consistent with the result of the expert evaluation.• Rank all alternatives: apply the model to rank all the alternatives.