2023. No. 3 Vol 17

      This article studies the predictive abilities of the generative-adversarial neural network approach in relation to time series using the example of price forecasting for the nodes of the Russian free electricity market for the day ahead. As a result of a series of experiments, we came to the conclusion that a generative adversarial network, consisting of two models (generator and discriminator), allows one to achieve a minimum of the error function with a greater generalizing ability than, all other things being equal, is achieved as a result of optimizing the static analogue of the generative model – recurrent neural network. Our own empirical results show that with a near-zero mean square error on the training set, which is demonstrated simultaneously by the recurrent and generative models, the error of the latter on the test set is lower. The adversarial approach also outperformed alternative reference models in out-of-sample forecasting accuracy: a convolutional neural network adapted for time series forecasting and an autoregressive linear model. Application of the proposed approach has shown that a generative-adversarial model with a given universal architecture and a limited number of explanatory factors, subject to additional training on data specific to the target node of the power system, can be used to predict prices in market nodes for the day ahead without significant deviations.

      In modern conditions, managerial decision-making is carried out using automated systems under the general name “Decision Support Systems” (DSS). When creating them, it is important to consider two key points. The first is the algorithmic component, which reflects the logic of the system as a whole and its individual parts. The second is the application interface through which the user interacts with it. The interface is a graphical interpretation of the algorithms that are implemented within the system. Therefore, it is very important to design and create such a relationship between the algorithm and the interface so that the user is as comfortable as possible using the DSS to solve current tasks (information input, its processing, presentation and analysis for decision making). Thus, there is a directly proportional relationship between the interface and the algorithm. Moreover, despite the fact that there are many studies on these aspects, both theoretical and practical, there are still questions to which one should pay attention to in terms of application. The purpose of this study is to formulate practical recommendations to prevent the entry of incorrect information into the DSS database and to present the results in a form convenient for its analysis. The main tasks of the work are to show by means of examples which errors can contribute to the entry of unreliable information into the database, as well as how best to present information on the monitor screen in accordance with the psychophysiological characteristics of a person in order to reduce the time for its analysis and decision-making.

      An important feature when working with financial data is the fact that the residuals of GARCH-models often have fatter tails than the tails of a normal distribution due to the large number of “outliers” in the data. This requires more detailed study. Kurtosis and quantile-based measure of heavy-tailedness were analyzed and compared in the article in relation to the problem of choosing the GARCH(1,1)-model specification. The data of indices of the Moscow Exchange were considered for the period from April 01, 2019 to February 22, 2022. Kurtosis values ​​ranged from 3 to 52. Empirical data showed that kurtosis was very sensitive to “outliers” in the data, which made it difficult to make assumptions about the distribution of model residuals. The approach considered in this paper based on the heavy-tailedness measure made it possible to justify the choice of degrees of freedom of the t-distribution for the model residuals to explain the fat tails in financial data. It was found that GARCH(1,1)-models with t(5)-distribution in the residuals are common.

      The development of intellectual capital theory through the introduction of the concept of implicitness involves considering intellectual capital as an implicit factor, so that the process of its formation is largely determined by the impact of specific hidden factors whose impact is expressed implicitly and is difficult to formalize. Currently, the process of selecting explicit and implicit factors affecting intellectual capital is not formalized in domestic and foreign studies, and therein is the relevance of this work. The purpose of this study was to develop a scheme for selecting explicit and implicit factors in the development of the organization’s intellectual capital in conjunction with its strategy based on a modified Balanced Scorecard, taking into account the distribution of indicators by types of cognitive activity. The implementation of this scheme was carried out by developing a fuzzy economic and mathematical model suitable for practical use. The main feature of the model is the possibility of fuzzy setting of “cut-off boundaries” for explicit and implicit factors. We present the results of testing the model on the example of a large regional university. Sets of explicit and implicit factors of the university’s intellectual capital are given for various “cut-off boundaries” using various defuzzification methods.

      This article presents a new simulation model of an intelligent transportation system (ITS) for the “smart city” with adaptive traffic light control. The proposed transportation model, implemented in the AnyLogic, allows us to study the behavior of interacting agents: vehicles (V) and pedestrians (P) within the framework of a multi-agent ITS of the “Manhattan Lattice” type. The spatial dynamics of agents in such an ITS is described using the systems of finite-difference equations with the variable structure, considering the controlling impact of the “smart traffic lights.” Various methods of traffic light control aimed at maximizing the total traffic of the ITS output flow have been studied, in particular, by forming the required duration phases with the use of a genetic optimization algorithm, with a local (“weakly adaptive”) switching control and based on the proposed fuzzy clustering algorithm. The possibilities of optimizing the characteristics of systems for individual control of the behavior of traffic lights under various scenarios, in particular, for the ITS with spatially homogeneous and periodic characteristics, are investigated. To determine the best values of individual parameters of traffic light control systems, such as the phases’ durations, the radius of observation of traffic and pedestrian flows, threshold coefficients, the number of clusters, etc., the previously proposed parallel real-coded genetic optimization algorithm (RCGA type) is used. The proposed method of adaptive control of traffic lights based on fuzzy clustering demonstrates greater efficiency in comparison with the known methods of collective impact and local (“weakly adaptive”) control. The results of the work can be considered a component of the decision-making system in the management of urban services.

      Financial time series are big arrays of information on quotes and trading volumes of shares, currencies and other exchange and over-the-counter instruments. The analysis and forecasting of such series has always been of particular interest for both research analysts and practicing investors. However, financial time series have their own features, which do not allow one to choose the only correct and well-functioning forecasting method. Currently, machine-learning algorithms allow one to analyze large amounts of data and test the resulting models. Modern technologies enable testing and applying complex forecasting methods that require volumetric calculations. They make it possible to develop the mathematical basis of forecasting, to combine different approaches into a single method. An example of such a modern approach is the Singular Spectrum Analysis (SSA), which combines the decomposition of a time series into a sum of time series, principal component analysis and recurrent forecasting. The purpose of this work is to analyze the possibility of applying SSA to financial time series. The SSA method was considered in comparison with other common methods for forecasting financial time series: ARIMA, Fourier transform and recurrent neural network. To implement the methods, a software algorithm in the Python language was developed. The method was also tested on the time series of quotes of Russian and American stocks, currencies and cryptocurrencies.