Algorithm of the morphological method of expert estimates for solving the forecasting problem
Abstract
The algorithm of one of the methods of forecasting the technical and economic indicators of an object of equipment or a sample of industry under conditions of uncertainty is given. The method is aimed at finding qualitatively stable states of the analyzed system. The main distinctive features of the method are: the ability to process the results of a survey of experts in the presence of quantitative and qualitative indicators, the possibility of making an informed decision with inconsistent answers from experts, the ability to choose a solution without calculating the efficiency indicator and the ability to assess the proximity of alternative solutions. The results of the expert survey are processed using a discrete probability space, the set of elementary events of which is a linear space of binary vectors. For each tuple of indicator estimates (x1,x2,...,xN ) injectively corresponds to a vector (y1, y2,..., yL) of the introduced linear space, where L >= N. At the same time, the range of acceptable values of each indicator xi corresponds to its own group of consecutive bits in the vectors of this space, where each bit represents a qualitatively homogeneous range of values of the indicator. To be able to compare two expert opinions in the space of binary vectors, the distance is determined, which is assumed to be equal to the number of qualitatively different indicators in the corresponding tuples of expert assessments. As a consequence, the criterion of qualitative homogeneity of two tuples of expert assessments is assumed to be the equality of the distance between the corresponding binary vectors to zero. An alternative forecast is made for each area of condensation of tuples of expert estimates. It is represented by averaging the values of each indicator in expert assessments that do not contradict each other. The choice of the final solution is based on the solution of a system of logical equations representing a set of areas of agreed expert assessments and restrictions on the values of indicators. In order to justify the decision, the experts who formed it are invited to explain their point of view.
References
E. A. Kharchenko, “The morphological approach to making reasonable decisions based on expert judgements,” Vestnik TvGU. Seriya: Prikladnaya Matematika, no. 2, pp. 42–56, 2019 (in Russian); doi:10.26456/vtpmk531
E. A. Kharchenko, “The general mechanical model of expert assessment methods,” in Sovremennye Tendencii Razvitiya Nauki i Obrazovaniya: Teoriya i Praktika, vol. 4, pp. 366–371, 2019 (in Russian).
A. V. Chugunov, Teoriya Elektronnyh Vychislitel’nyh Mashin i Vychislitel’nyh Sistem [Theory of Electronic Computers and Computing Systems], Moscow: Ministry of Defense of the USSR, 1990 (in Russian).
O. A. Mikhaleva, Matematicheskoe i Programmnoe Obespechenie Obrabotki Rezul’tatov Gruppovogo Ocenivaniya dlya Upravleniya Setevoj Ekspertizoj v Raspredelennoj Srede [Mathematical and Software for Processing Group Evaluation Results for Network Expertise Management in a Distributed Environment], Bryansk, Russia: Bryanskii gosudarstvennyi tekhnicheskii universitet, 2020 (in Russian).
E. A. Burkov, Metody i Algoritmy Analiza i Agregirovaniya Gruppovyh Ekspertnyh Ocenok [Methods and Algorithms for Analysis and Aggregation of Group Expert Evaluations], Saint Petersburg, Russia: Saint Petersburg Electrotechnical University, 2011 (in Russian).
G. M. Gambarov, N. M. Zhuravel, and Yu. G. Korolev, Statisticheskoe Modelirovanie i Prognozirovanie [Statistical Modeling & Forecasting], Moscow: Ministry of Defense of the USSR, 1990 (in Russian).
I. V. Bestuzhev-Lada, Rabochaya kniga po prognozirovaniyu [Forecasting workbook], Moscow: Mysl’, 1982 (in Russian).
E. A. Kharchenko, “Morphological Approach to Anomaly Detection in Multidimensional Data,” in Matematika: teoreticheskie i prikladnye issledovaniya, Moscow: Moskovskii Politekh, pp. 194–198, 2022 (in Russian).
This work is licensed under a Creative Commons Attribution 4.0 International License.
