基于決策者風(fēng)險(xiǎn)偏好大數(shù)據(jù)分析的大群體應(yīng)急決策方法

摘要： 針對(duì)重大突發(fā)事件應(yīng)急決策大群體成員的風(fēng)險(xiǎn)偏好復(fù)雜難測(cè)問(wèn)題,提出了一種新的基于決策者風(fēng)險(xiǎn)偏好大數(shù)據(jù)分析的大群體應(yīng)急決策方法。首先專(zhuān)家群體對(duì)突發(fā)事件進(jìn)行快速響應(yīng),生成若干應(yīng)急預(yù)案及其風(fēng)險(xiǎn)屬性信息;其次,社會(huì)公眾通過(guò)網(wǎng)絡(luò)等渠道參與到應(yīng)急決策中來(lái)并形成決策大群體,給出不同預(yù)案的偏好值;然后,利用證據(jù)推理算法得出公眾對(duì)各預(yù)案的風(fēng)險(xiǎn)效用值,將預(yù)案風(fēng)險(xiǎn)效用值與預(yù)案偏好值加權(quán)組合,得到各個(gè)預(yù)案的大群體決策者的風(fēng)險(xiǎn)偏好值;最后,基于風(fēng)險(xiǎn)偏好值,利用大數(shù)據(jù)分析技術(shù)對(duì)大群體的風(fēng)險(xiǎn)偏好進(jìn)行聚類(lèi)識(shí)別,從中篩選出風(fēng)險(xiǎn)中立者組成新的應(yīng)急決策群體,再次聚類(lèi)得出應(yīng)急決策群體的成員組成結(jié)構(gòu),以此為基礎(chǔ)計(jì)算決策者權(quán)重和應(yīng)急預(yù)案的最終效用值,得應(yīng)急預(yù)案排序結(jié)果。最后通過(guò)算例分析驗(yàn)證了方法的有效性和可行性。
According to the complex and unpredictable risk-preference of emergency decision group members, a new large group emergency decision method based on large data analysis of risk preference of decision-makers is proposed in this paper. First, the expert group responds quickly to the emergency, and generates some contingency plans and their risk attributes information. Secondly, through the network and other channels the public participates in emergency decision-making and form a large group of decision-making, giving different preplans preference values. Then, this paper uses evidential reasoning algorithm to get the public's risk utility value for each plan, and combines the preplan risk utility value with the preplan preference value to get the risk preference of large group decision makers of each plan. Finally, based on the risk preference value, this paper uses the big data analysis technology to identify the risk preferences of large groups, and selects the risk neutral group to form a new emergency decision-making group. The structure of the members of the emergency decision group can be obtained by reclustering and on this basis, the decision maker's weight and the final utility value of the emergency plan are calculated, and thus the order results of emergency plan also are obtained. Finally, the effectiveness and feasibility of the method are verified by an example analysis.