Multi-criteria decision analysis
Multi-criteria decision analysis (MCDA) is a collection of methods and procedures for selecting, comparing, and evaluating alternatives when decisions must be made based on several, often competing, criteria. This approach reflects the complexity of real-world problems where it is impossible to reduce the choice to a single performance indicator.
Essence and Goals
Multi-criteria situations arise when various aspects of a solution's quality (e.g., cost, reliability, efficiency, sustainability) are significant to the decision-maker (DM) and require simultaneous consideration. The main goal of multi-criteria analysis is to find solutions that best satisfy all criteria or represent an optimal compromise among them.
Advantages and Limitations
Advantages of the multi-criteria approach:
- Consideration of various aspects of the problem situation;
- More flexible expression of the decision-maker's preferences;
- Increased transparency in the decision-making process;
- Ability to consider the interests of various stakeholders.
Limitations and challenges:
- Difficulty in formalizing preferences and criteria weights;
- Potential conflicts between criteria;
- Incomparability of some alternatives;
- Dependence of results on subjective assessments and scale structures.
Basic Principles
- Decomposition of overall quality into specific criteria: allows for a more accurate reflection of the preference structure and the situation being analyzed.
- Formation of a set of efficient (Pareto-optimal) solutions: used when it is not possible to aggregate all criteria in a single way.
- Compensation and weighting coefficients: assigned by the decision-maker or calculated according to formal rules, reflecting the relative importance of the criteria.
- Aggregation of criteria: transforming a multi-criteria problem into an equivalent single-criterion one using an aggregating function.
- Visualization methods: used to display achievable combinations of criteria values and facilitate choice.
Classification of Methods
Multi-criteria analysis methods are divided into several groups depending on the type of information and the form of preferences:
- Methods based on quantitative indicators with numerical utility values;
- Methods that convert qualitative assessments into numerical ones;
- Comparison methods without calculating a numerical value;
- Verbal analysis methods using linguistic judgments.
Examples of Methods
- Analytic Hierarchy Process (AHP);
- Outranking methods (ELECTRE, PROMETHEE);
- Methods with additive or multiplicative aggregation;
- Human-machine iterative methods;
- Methods based on choice functions.