Hybrid Intelligent System for Forecasting the Russian Economy’s Need for Qualified Personnel

Abstract

The article discusses a hybrid intelligent system for forecasting the Russian economy's demand for skilled personnel, with an emphasis on the analysis of time series of vacancies. In a dynamically changing labor market, accurate forecasts allow employers and job seekers to plan their strategies more effectively. The study covers the period from June 2022 to November 2023 and uses data from leading job sites. The work uses both classical statistical methods and modern machine learning technologies, including Singular Spectrum Analysis (SSA), XGBoostRegressor, and an innovative TimeLLM approach based on Large Language Models (LLM). The hybrid approach combining SSA and XGBoostRegressor, as well as the TimeLLM method, showed high forecasting accuracy, especially for IT specialties and executive positions. The quality of forecasts was assessed using the Mean Absolute Error and Mean Absolute Percentage Error metrics. The results demonstrate the advantages of hybrid and LLM methods over classical approaches, although they point to possible limitations associated with going beyond the permissible values. It is noted that hybrid approaches and LLM-based methods allow achieving high predictive accuracy, but require taking into account the specificity of the data and possible correction of extreme values. It is recommended to use TimeLLM as a flexible tool capable of integrating exogenous variables and adjusting to specific forecasting tasks

Please cite this article in English as:

Proletarsky A.V., Berezkin D.V., Tsaplin S.T. Hybrid intelligent system for forecasting the Russian economy's need for qualified personnel. Herald of the Bauman Moscow State Technical University, Series Instrument Engineering, 2026, no. 2 (155), pp. 80--91 (in Russ.). EDN: CUNSTQ

References

[1] Makridakis S., Spiliotis E., Assimakopoulos V. The M4 Competition: 100,000 time series and 61 forecasting methods. Int. J. Forecast., 2020, vol. 36, no. 1, pp. 54--71. DOI: https://doi.org/10.1016/j.ijforecast.2019.04.014

[2] Hyndman R.J., Athanasopoulos G. Forecasting: principles and practice. Melbourne, OTexts, 2018.

[3] Lim B., Arık S.O., Loeff N., et al. Time-series forecasting with deep learning: a survey. Philos. Trans. R. Soc. A, 2021, vol. 379, no. 2194, art. 20200209. DOI: https://doi.org/10.1098/rsta.2020.0209

[4] Hamilton J.D. Time series analysis. Princeton, Princeton University Press, 1994.

[5] Box G.E.P., Jenkins G.M., Reinsel G.C., et al. Time series analysis. Hoboken, John Wiley & Sons, 2015.

[6] Shumway R.H., Stoffer D.S. Time series analysis and its applications. New York, Springer, 2017.

[7] Diebold F.X. Elements of forecasting. Cincinnati, South-Western, 2001.

[8] Harvey A.C. Forecasting, structural time series models and the Kalman filter. Cambridge, Cambridge University Press, 1990.

[9] Cowpertwait P.S.P., Metcalfe A.V. Introductory time series with R. New York, Springer, 2009.

[10] Bengio Y., Goodfellow I., Courville A. Deep learning. Cambridge, MIT Press, 2016.

[11] Hochreiter S., Schmidhuber J. Long short-term memory. Neural Comput., 1997, vol. 9, no. 8, pp. 1735--1780. DOI: https://doi.org/10.1162/neco.1997.9.8.1735

[12] Vaswani A., Shazeer N., Parmar N., et al. Attention is all you need. NIPS, 2017, vol. 30. URL: https://arxiv.org/pdf/1706.03762

[13] Kingma D.P., Ba J. Adam: a method for stochastic optimization. arXiv:1412.6980. DOI: https://doi.org/10.48550/arXiv.1412.6980

[14] Cleveland R.B., Cleveland W.S., McRae J.E., et al. STL: a seasonal-trend decomposition procedure based on Loess. Journal of Official Statistics, 1990, vol. 6, no. 1, pp. 3--73.

[15] Brownlee J. Machine learning mastery with Python. Vermont, Machine Learning Mastery, 2016.

[16] Seni G., Elder J.F. Ensemble methods in data mining. Berlin, Springer, 2010.

[17] Dietterich T.G. Ensemble methods in machine learning. In: Multiple classifier systems. Springer, 2000, pp. 1--15. DOI: https://doi.org/10.1007/3-540-45014-9_1

[18] Zhou Z.H. Ensemble methods. Foundations and algorithms. London, CRC Press, 2012.

[19] Jin M., Shiyu W., Ma L., et al. Time-LLM: time series forecasting by reprogramming large language models. arXiv:2310.03506. DOI: https://doi.org/10.48550/arXiv.2310.01728

[20] Laptev N., Yosinski J., Lippi M., et al. Time series extreme event forecasting with neural networks at uber. In: International Conference on Machine Learning, 2017, vol. 34, pp. 1--5.