sciresНаучные исследования экономического факультета. Электронный журналScientific Research of Faculty of Economics. Electronic Journal2078-3809Moscow State University10.38050/2078-3809-2025-17-4-9-34scires-421Research ArticleВОПРОСЫ ТЕОРИИTHEORETICAL ISSUESИнновационные подходы к генерации обучающих данных для прогнозирования спроса на нефтьInnovative Approaches to Training Data Generation for Oil Demand ForecastingМанаховаИ. В.ManakhovaI. V.

Манахова Ирина Викторовна, доктор экономических наук, профессор, экономический факультет

г. Москва

Irina V. Manakhova, Doctor of Economics, Professor, Faculty of Economics

Moscow 

manakhovaiv@mail.ruМатыцынА. В.MatytsynA. V.

Матыцын Александр Владимирович, магистр; соискатель ученой степени кандидата экономических наук, экономический факультет

г. Париж

г. Москва

Aleksandr V. Matytsyn, Master’s Degree; Candidate of Sciences Degree in Economics, Faculty of Economics

Paris

Moscow

avmatytsyn@gmail.comМГУ имени М.В. ЛомоносоваРоссияLomonosov Moscow State UniversityRussian FederationВысшая школа внешней торговли; МГУ имени М.В. ЛомоносоваФранцияÉcole Supérieure du Commerce Extérieur (ESCE); Lomonosov Moscow State UniversityFrance202505042026174934Copyright © Манахова И.В., Матыцын А.В., 20262026Манахова И.В., Матыцын А.В.Manakhova I.V., Matytsyn A.V.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://scires.elpub.ru/jour/article/view/421

В данной работе исследуются методы генерации обучающих данных для повышения точности прогнозирования спроса на рынке нефти. Рассматриваются ограничения традиционных подходов и обосновывается применение генеративно-состязательных сетей, в частности модели TimeGAN (Time-series Generative Adversarial Network), для создания синтетических временных рядов. Результаты показывают, что TimeGAN позволяет генерировать реалистичные данные, приближенные к реальным, с сохранением волатильности и структурных особенностей рынка. Также выявлены ограничения модели, требующие дальнейшего исследования для повышения эффективности и точности прогнозирования спроса на нефть в условиях рыночной нестабильности.

This study explores methods for generating training data to improve demand forecasting accuracy in the oil market. The limitations of traditional approaches are examined, and the use of generative adversarial networks, specifically the TimeGAN (Time-series Generative Adversarial Network) model, is proposed for creating synthetic time series data. The results demonstrate that TimeGAN can generate realistic data closely resembling actual data, preserving market volatility and structural characteristics. However, model limitations were identified, suggesting the need for further research to enhance forecast efficiency and accuracy on the oil in volatile market conditions.

глубокое обучениегенеративно-состязательные сетиобучающие данныемодель TimeGAN.deep learninggenerative adversarial networkstraining datamodel TimeGAN.ReferencesКаукин А.С., Павлов П.Н., Косарев В.С. Краткосрочное прогнозирование цен на электроэнергию с использованием генеративных нейронных сетей // Бизнес-информатика. 2023. № 3 (17). С. 7–23.Kaukin A.S., Pavlov P.N., Kosarev V.S. Kratkosrochnoe prognozirovanie tsen na elektroenergiyu s ispol'zovaniem generativnykh neyronnykh setey. Biznes-informatika. 2023. No. 3 (17). P. 7–23. (In Russ.).Копытин И.А. Трансформация рынка нефти в Европе: тенденции и перспективы // Проблемы прогнозирования. 2024. № 6. С. 217–226. https://doi.org/10.47711/0868-6351-207-217-226.Kopytin I.A. Transformatsiya rynka nefti v Evrope: tendentsii i perspektivy. Problemy prognozirovaniya. 2024. No. 6. P. 217–226. https://doi.org/10.47711/0868-6351-207-217-226. (In Russ.).Рабчевский А.Н. Обзор методов и систем генерации синтетических обучающих данных // Прикладная математика и вопросы управления. 2023. № 4. С. 6–45. https://doi.org/10.15593/2499-9873/2023.4.01.Rabchevskiy A.N. Obzor metodov i sistem generatsii sinteticheskikh obuchayushchikh dannykh. Prikladnaya matematika i voprosy upravleniya. 2023. No. 4. P. 6–45. https://doi.org/10.15593/2499-9873/2023.4.01. (In Russ.).Розенцвайг А.К. Методы эконометрического моделирования и анализа социально-экономических явлений. Набережные Челны, 2014. 121 с. https://doi.org/10.13140/RG.2.1.3998.5526.Rozentsvayg A.K. Metody ekonometricheskogo modelirovaniya i analiza sotsial'noekonomicheskikh yavleniy. Naberezhnye Chelny, 2014. 121 p. https://doi.org/10.13140/RG.2.1.3998.5526. (In Russ.).Brajard J., Carrassi A., Bocquet M., Bertino L.. Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: A case study with the Lorenz 96 model // Journal of Computational Science. 2020. Vol. 44. https://doi.org/101171.10.1016/j.jocs.2020.101171.Brajard J., Carrassi A., Bocquet M., Bertino L.. Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: A case study with the Lorenz 96 model. Journal of Computational Science. 2020. Vol. 44. https://doi.org/101171.10.1016/j.jocs.2020.101171.Farzana G.S., Prakash N. Machine Learning in Demand Forecasting // A Review. In Proceedings of the 2nd International Conference on IoT, Social, Mobile, Analytics and Cloud in Computational Vision and Bio-Engineering (ISMAC-CVB 2020). 2020. http://dx.doi.org/10.2139/ssrn.3733548.Farzana G.S., Prakash N. Machine Learning in Demand Forecasting. A Review. In Proceedings of the 2nd International Conference on IoT, Social, Mobile, Analytics and Cloud in Computational Vision and Bio-Engineering (ISMAC-CVB 2020). 2020. http://dx.doi.org/10.2139/ssrn.3733548.Juneja T., Bajaj S., Sethi N. Synthetic Time Series Data Generation Using Time GAN with Synthetic and Real-Time Data Analysis // Proceedings of the International Conference on Advances in Computing and Data Sciences. Springer. 2023. P. 567–576. https://doi.org/10.1007/978-981-99-0601-7_51.Juneja T., Bajaj S., Sethi N. Synthetic Time Series Data Generation Using Time GAN with Synthetic and Real-Time Data Analysis. Proceedings of the International Conference on Advances in Computing and Data Sciences. Springer. 2023. P. 567–576. https://doi.org/10.1007/978-981-99-0601-7_51.Kanagarathinam K. Comprehensive Overview of Optimization Techniques in Machine Learning Training // Computational Science and Optimization Letters. 2024. Vol. 2. No. 1. P. 69–85. https://doi.org/10.59247/csol.v2i1.69.Kanagarathinam K. Comprehensive Overview of Optimization Techniques in Machine Learning Training. Computational Science and Optimization Letters. 2024. Vol. 2. No. 1. P. 69–85. https://doi.org/10.59247/csol.v2i1.69.Li J., Liu Y., Li Q. Generative Adversarial Network and Transfer Learning Based Fault Detection for Rotating Machinery with Imbalance Data Condition // Measurement Science and Technology. 2022. https://doi.org/33.10.1088/1361-6501/ac3945.Li J., Liu Y., Li Q. Generative Adversarial Network and Transfer Learning Based Fault Detection for Rotating Machinery with Imbalance Data Condition. Measurement Science and Technology. 2022. https://doi.org/33.10.1088/1361-6501/ac3945.Liu Y., Liang Z., Li X. Enhancing Short-Term Power Load Forecasting for Industrial and Commercial Buildings: A Hybrid Approach Using TimeGAN, CNN, and LSTM // IEEE Open Journal of the Industrial Electronics Society. 2023. No. 4. P. 451–462. https://doi.org/10.1109/OJIES.2023.3319040.Liu Y., Liang Z., Li X. Enhancing Short-Term Power Load Forecasting for Industrial and Commercial Buildings: A Hybrid Approach Using TimeGAN, CNN, and LSTM. IEEE Open Journal of the Industrial Electronics Society. 2023. No. 4. P. 451–462. https://doi.org/10.1109/OJIES.2023.3319040.Matta C., Bianchesi N.M., Oliveira M., Balestrassi P., Leal F. A comparative study of forecasting methods using real-life econometric series data // Production. 2021. Vol. 31. https://doi.org/10.1590/0103-6513.20210043.Matta C., Bianchesi N.M., Oliveira M., Balestrassi P., Leal F. A comparative study of forecasting methods using real-life econometric series data. Production. 2021. Vol. 31. https://doi.org/10.1590/0103-6513.20210043.Moroff N., Kurt E., Kamphues J. Machine Learning and Statistics: A Study for Assessing Innovative Demand Forecasting Models // Procedia Computer Science. 2021. No.180. P. 40–49. https://doi.org/10.1016/j.procs.2021.01.127.Moroff N., Kurt E., Kamphues J. Machine Learning and Statistics: A Study for Assessing Innovative Demand Forecasting Models. Procedia Computer Science. 2021. No.180. P. 40–49. https://doi.org/10.1016/j.procs.2021.01.127.Mushunje L., Allen D., Peiris S. Volatility and Irregularity Capturing in Stock Price Indices Using Time Series Generative Adversarial Networks (TimeGAN) // Computational Engineering, Finance, and Science. 2023. https://doi.org/10.48550/arXiv.2311.12987.Mushunje L., Allen D., Peiris S. Volatility and Irregularity Capturing in Stock Price Indices Using Time Series Generative Adversarial Networks (TimeGAN). Computational Engineering, Finance, and Science. 2023. https://doi.org/10.48550/arXiv.2311.12987.Reboredo J.C., Ugolini A. Quantile dependence of oil price movements and stock returns // Energy Economics. 2016. Vol. 54. P. 33–49. https://doi.org/10.1016/j.eneco.2015.11.015.Reboredo J.C., Ugolini A. Quantile dependence of oil price movements and stock returns. Energy Economics. 2016. Vol. 54. P. 33–49. https://doi.org/10.1016/j.eneco.2015.11.015.Sacco M.A., Ruiz J.J., Pulido M., Tandeo P. Evaluation of Machine Learning Techniques for Forecast Uncertainty Quantification // Quarterly Journal of the Royal Meteorological Society. 2021. Vol. 147. No. 34. P. 1234–1251. https://doi.org/10.48550/arXiv.2111.14844.Sacco M.A., Ruiz J.J., Pulido M., Tandeo P. Evaluation of Machine Learning Techniques for Forecast Uncertainty Quantification. Quarterly Journal of the Royal Meteorological Society. 2021. Vol. 147. No. 34. P. 1234–1251. https://doi.org/10.48550/arXiv.2111.14844.Yue T, Liu Y. Multi-Scale Price Forecasting Based on Data Augmentation // Applied Sciences. 2024. Vol. 14. No. 19. P. 8737. https://doi.org/10.3390/app14198737.Yue T, Liu Y. Multi-Scale Price Forecasting Based on Data Augmentation. Applied Sciences. 2024. Vol. 14. No. 19. P. 8737. https://doi.org/10.3390/app14198737.

The authors declare that there are no conflicts of interest present.