SECA (Simultaneous Evaluation of Criteria and Alternatives)

SECA is a method that ranks alternatives by a single score once you have a numerical decision table. Most ranking methods perform two separate jobs in sequence: weights are first found by some method (AHP, Entropy, CRITIC…), and the alternatives are then ranked with those weights. SECA solves both jobs together, in a single optimisation step. This is where its name comes from: the simultaneous evaluation of criteria and alternatives. Its output is a score for every alternative and the ranking that score produces; criterion weights also emerge from within the same calculation. It was proposed by Keshavarz Ghorabaee, Amiri, Zavadskas, Turskis and Antucheviciene in 2018.

The idea behind SECA is this: a criterion's weight is related to how discriminating that criterion is within the decision table, but looking at discriminating power alone is not enough, because the weights must also carry the alternatives to a good total score. SECA attempts both at once. On one hand it seeks a weight set that does not overly disregard the criteria's dispersion (their variance); on the other, it wants the alternatives' total score, under those same weights, to come out high. It combines the two aims with a balancing coefficient called β; as β grows, more weight goes to the total score, and as it shrinks, more weight goes to balance among the criteria.

This idea carries a philosophical consequence: because SECA combines objective weighting and ranking in the same step, the weights are not "an external judgement" but a result arising from the data itself. Where the decision-maker has no particular order of priority, or wants to see that order emerge from the data, SECA is the right place; where a strong expert opinion on the criteria's importance already exists, there is no direct way of feeding that opinion into SECA, because the weights are the output of the calculation, not its input.

The method proceeds through three steps.

First, scale equalisation. Criteria are expressed in different units. SECA spreads every column between its own smallest and largest value onto a 0 to 1 scale: the worst value becomes 0, the best becomes 1. For "less is better" criteria such as cost, this direction is reversed. As a result, all columns sit on the same scale and become comparable with one another.

Second, finding weight and score together. This is SECA's distinguishing step. The method sets up an optimisation problem: it searches for criterion weights that both raise the alternatives' total score and limit imbalance among the criteria, that is, prevent any one criterion's weight from becoming excessively dominant. These two aims are weighted by the coefficient β and solved as a single numerical optimisation problem. The solution's output is both the criterion weights and every alternative's total score; the two emerge from the same calculation at the same time.

Third, ranking. SECA arranges the alternatives from highest to lowest by the total score obtained.

DecisionMind solves this optimisation problem with a numerical solver, trying multiple starting points, and holds β fixed at its default value of 0.5; a user wanting a different balance can change β. The formulas behind each step, the intermediate tables and citation formats are given on the DecisionMind method page; this card carries no formulas.

The score SECA gives shows how good a total result an alternative achieves, relative to the other alternatives in this table, under weights derived from the data. The score alone is not a "percentage success" and cannot be compared with a score from a different analysis, because both the normalisation and the weights are specific to this alternative set. The resulting weights should be read in the same way: a high weight for a criterion does not mean "the decision-maker cares about this criterion," it means "in this data set, this criterion's power to discriminate among the alternatives was found to be high and it raised the total score."

Thus instead of writing:

"SECA revealed the criteria's true importance"

the report should read:

"For this alternative set and this β value, the weights derived from the data are as follows; the alternative with the highest total score is this one"

SECA works with crisp data: a single number in every cell. DecisionMind currently holds no fuzzy, grey or other data-type extension of SECA; the method is offered in a single form.

You need alternatives in rows, criteria in columns, a number in every cell and no empty cells; direction information ("more is better" or "less is better") is required for every criterion. No external weight is required, and even if one is entered, SECA does not use it; the weights come from the calculation itself. A minimum of two alternatives and two criteria is sufficient, and three to twelve criteria work comfortably. SECA can be applied provided there is no strong dependency among the criteria (one does not automatically rise as another rises) and the table is fully populated.

If you have a numerical decision table and do not wish to fix weights in advance, or lack a reliable expert opinion to set them, SECA is a suitable choice. Its typical areas are supplier evaluation and alternative selection; it can be used wherever both weight and ranking are wanted straight from the data.

The case where it should not be used is this: if you already hold a sound, defensible expert opinion on the criteria's importance and want that opinion reflected in the decision, SECA is not suitable, because it does not accept weights from outside. In that case, weights should be set with a subjective method such as AHP, BWM or SWARA, and the result then passed to a ranking method that accepts weights, such as TOPSIS or VIKOR. If the criteria are strongly dependent on one another, that link should be resolved first.

A numerical table, no weight in hand, weights should be derived from the data → SECA

Weight should come from expert opinion → AHP, BWM, SWARA (subjective) for weight, then a ranking method such as TOPSIS/VIKOR

Weight wanted purely from the data's dispersion, independent of ranking → Entropy, CRITIC

The question of how much the results depend on weight changes → sensitivity analysis

SECA's most important advantage is that it removes the risk of inconsistency between weight-setting and ranking. Where the two are done in separate steps, the weighting method and the ranking method can carry assumptions that do not match one another; SECA reduces this risk by combining them in a single calculation. Its β coefficient also makes the trade-off between "balance among the criteria" and "raising the total score" an explicitly adjustable parameter, a flexibility that can be adapted to different decision cultures.

Its limitations are specific to the method. First, SECA's solution rests on a nonlinear optimisation problem; the solution to such problems can be sensitive to the starting point, and different solvers or starting values can give weights that are close to but not exactly the same as one another (Keshavarz Ghorabaee et al., 2018). Second, the choice of β directly affects the result; there is no single "correct" value for β, it is a decision preference and must be justified in the report. Third, because the weights are derived from the data, there is no direct way of bringing expert knowledge into the decision. Fourth, with a very small number of alternatives or criteria, it can become difficult for the weights to separate meaningfully from the data.

The most common mistake is presenting the weights SECA produces as though they were the decision-maker's own subjective order of priority. These weights are a property of the data set, not anyone's opinion, and the report must clearly separate the two.

A second mistake is leaving the β value at its default without ever questioning it and presenting the result as a single "correct answer"; whether the ranking changes when β is changed should be checked. A third mistake is entering an external weight set and assuming SECA will use it; SECA ignores any weights entered. A fourth mistake is accepting the optimisation solution as final without checking that it has converged; it should be resolved from different starting points and the scores checked for stability.

The governing principle is this:

The weight and score SECA gives are a result of the chosen alternative set and the β value; they do not substitute for expert opinion, they offer a data-based starting point.