Evaluation of alternatives
Evaluation of alternatives is a key stage in the decision-making process where the decision-maker (DM) identifies and expresses their preferences for possible alternatives by comparing them based on significant attributes and characteristics. This stage precedes the final choice and helps establish the basis for making a well-founded decision.
The evaluation of alternatives relies on a system of criteria and scales. Scales allow for the formalization of the features of the alternatives, while criteria reflect the important aspects by which they are compared. Depending on the situation, various types of scales are used: from simple ones (e.g., classification or ranking) to more complex ones (e.g., assessing relative attractiveness).
Classification of Methods for Evaluating Alternatives
Methods for evaluating alternatives are tools that allow for the formalization of the decision-maker's (DM) preferences, ordering or ranking the alternatives, identifying the best ones, or determining acceptable ones. These methods differ based on the type of information used, the form of preference expression, the degree of formalization, the number of criteria, and other factors.
1. By Type of Information Used
- Quantitative methods are used when the characteristics of alternatives are represented numerically. These methods are based on measurable parameters such as cost, time, and performance. They allow for the use of interval and ratio scales, as well as calculations (e.g., weighted sums, normalization, aggregation).
- Qualitative methods are used when information is presented in a verbal or subjective form. Nominal and ordinal scales are utilized, and the assessments themselves are given through expert judgments, surveys, or linguistic scales (e.g., "high level," "satisfactory," "low risk").
- Mixed methods combine numerical and verbal information. They are often used in multi-criteria evaluation where some criteria are quantitative and others are qualitative. For example, the Analytic Hierarchy Process (AHP) allows for the conversion of verbal assessments into numerical form through pairwise comparisons.
2. By Form of DM Preference Expression
- Ranking methods involve ordering all alternatives by preference without specifying how much better one is than another.
- Pairwise comparison methods allow preferences to be identified through the one-on-one comparison of alternatives. They are used in both qualitative and quantitative forms. The method may be based on a comparison matrix that reflects which of two alternatives is preferred and by how much.
- Scale-based evaluation methods use numerical or symbolic scales to express the degree to which characteristics are present. The choice of scale type (nominal, ordinal, interval, ratio) depends on the nature of the criterion.
- Single-criterion (aggregation) methods reduce the characteristics of alternatives to a single generalized score—an objective function, utility function, value function, etc. They are typically used when it is necessary to select the optimal alternative based on an integral characteristic.
3. By Degree of Method Formalization
- Formalized methods rely on clear algorithms and procedures and can be automated. These include, for example, criteria aggregation methods, multi-criteria optimization, utility methods, and scale-based evaluation.
- Non-formalized methods involve intuitive or expert evaluation without strict computational procedures. They are used when the problem is poorly structured or when it is impossible to formalize preferences (e.g., the Delphi method, brainstorming, heuristic rules).
- Combined (hybrid) methods combine formal calculations with the ability to adjust assessments through interaction with the DM or experts. An example is interactive procedures, where the DM refines preferences as intermediate results are analyzed.
4. By Number of Criteria Evaluated
- Single-criterion methods assume that the choice is made based on a single, predetermined criterion. This may involve simple comparisons based on the value of a target indicator: price, efficiency, time, etc.
- Multi-criteria methods consider several criteria simultaneously, including those of different types (e.g., cost, reliability, quality). Such methods include procedures for normalization, weighting of criteria, aggregation, and compromise solutions.
5. By Degree of DM Participation in the Evaluation Process
- Automatic (axiomatic) methods do not require DM participation after the initial data is entered. The decision is generated based on predefined formal rules (e.g., a utility function or the logic of choice axioms).
- Interactive methods involve step-by-step participation from the DM: at each stage, they can evaluate the proposed alternatives, change the problem parameters, or clarify their preferences. These methods are particularly useful for multi-criteria choice.
- Hybrid methods combine automated procedures with the possibility of expert intervention. For example, in ranking or clustering methods, the result can be adjusted based on expert feedback.
6. By Degree of Information Certainty
- Methods under certainty assume that all information about the alternatives is known precisely and is not in doubt.
- Methods under risk account for the probabilistic nature of outcomes (e.g., expected value, criterion of maximum probability of success).
- Methods under uncertainty are used when precise information about consequences is unavailable, requiring the use of subjective assessments, scenario analyses, or expert hypotheses.