A method based on cloud model and FCM clustering for risky large group decision making

A risky large group decision-making method based on FCM clustering and cloud models is proposed for risky large group decision-making problems with linguistic evaluation scales, unknown attribute weights, and many decision members with unknown weights, considering the psychological behavioral characteristics of decision makers' regret avoidance. The method first uses the golden partition method to improve the cloud model to transform the uncertain linguistic evaluation matrix into a comprehensive cloud model, which quantifies the fuzziness and randomness of linguistic values. The cloud model expectation values are then extracted to determine the attribute weights using the entropy weighting method. Secondly, the three numerical features of the cloud model are extracted as sample features for FCM clustering to obtain the decision maker's preference clustering information, and the initial weights of decision-makers are determined according to the majority principle, which improves the existing studies that simply use the expected value of the cloud model for clustering analysis, ignoring the entropy and super entropy for portraying the ambiguity and randomness. On this basis, the Hamming distance is introduced to calculate the closeness to adjust the initial weights of decision-makers, improving the way that the weights of aggregation members are equally distributed in previous studies. Finally, considering the influence of the decision maker's psychological behavior on decision information in the risky decision-making process, regret theory is introduced to construct a decision maker's perceived utility matrix, which is combined with the decision maker's weights to determine and rank the combined perceived utility. Through comparison with existing methods, it is found that the proposed method of recalibration of decision-maker preference clustering, while considering the psychological behavior of decision-maker regret avoidance, not only solves the situation of large group decision making in which expert information is easily distorted but also satisfies the convenience of the calculation process and is more suitable for the situation where there are many decision-makers and their preferences are complicated.