Mathematical optimization for time series decomposition (original) (raw)

Abstract

Decomposing time series into trend and seasonality components reveals insights used in forecasting and anomaly detection. This study proposes a mathematical optimization approach that addresses several data-related issues in time series decomposition. Our approach does not only handle longer and multiple seasons but also identifies outliers and trend shifts. Numerical experiments on real-world and synthetic problem sets present the effectiveness of the proposed approach.

Access this article

Log in via an institution

Subscribe and save

• Starting from 10 chapters or articles per month
• Access and download chapters and articles from more than 300k books and 2,500 journals
• Cancel anytime View plans

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Time Series

Chapter © 2023

Notes

1. The user can purposefully set M value and limit trend shifts. We highlight even a large number does not lead to an unrealistic solution due to the rest of the constraints.
2. Similar to the trend shift, the user can set M value if spikes or dips are to be bounded.

References

• Arputhamary B, Lawrence DLA (2018) A pragmatic study on time series models for big data. Int J Emerg Res Manag Technol 6:67
  Article Google Scholar
• Bartholomew DJ (1971) Time series analysis forecasting and control. J Oper Res Soc 22(2):199–201
  Article Google Scholar
• Bleikh H, Young W (2016) Time series analysis and adjustment: measuring, modelling and forecasting for business and economics. Taylor & Francis, Thames
  Book Google Scholar
• Brockwell PJ, Davis RA editors (2002) Nonstationary and seasonal time series models. Springer, New York, pp 179–221
• Brockwell PJ, Davis RA, Calder MV (2002) Introduction to time series and forecasting, vol 2. Springer, Berlin
  Book Google Scholar
• Chen Z-G, Anderson OD (1998) Increment-vector methodology: transforming non-stationary series to stationary series. J Appl Prob 35(1):64–77
  Article Google Scholar
• Cleveland RB, Cleveland WS, McRae JE, Terpenning I (1990) STL: a seasonal-trend decomposition procedure based on loess (with discussion). J Off Stat 6:3–73
  Google Scholar
• Cleveland WS, Dunn DM, Terpenning IJ (1978) SABL: A resistant seasonal adjustment procedure with graphical methods for interpretation and diagnosis. In: National Bureau of Economic Research, Inc, June
• Dagum E (1980) The X-II-ARIMA seasonal adjustment method. In: Essays Collection of Estela Bee Dagum in Statistical Sciences, 01
• Dokumentov A, Hyndman RJ (2015) STR: A seasonal-trend decomposition procedure based on regression. In: Monash econometrics and business statistics working papers 13/15, Monash University, Department of Econometrics and Business Statistics
• Duarte FS, Rios RA, Hruschka ER, de Mello RF (2019) Decomposing time series into deterministic and stochastic influences: a survey. Digital Signal Process 95:102582
  Article Google Scholar
• Findley DF, Monsell BC, Bell WR, Otto MC, Chen B-C (1998) New capabilities and methods of the X-12-ARIMA seasonal-adjustment program. J Bus Econ Stat 16(2):127–152
  Google Scholar
• Gomes J, Velho L (2015) From Fourier analysis to wavelets, vol 01. Springer, Berlin
  Book Google Scholar
• Gurobi Optimization (2020) Gurobi optimizer reference manual,
• Hamilton JD (2017) Why you should never use the Hodrick–Prescott filter. In: Working paper 23429, National Bureau of Economic Research, May
• Hassani H (2007) Singular spectrum analysis: methodology and comparison. University Library of Munich, Germany, MPRA Paper, vol 5, p 01
• Hodrick RJ, Prescott E. (1981) Post-War U.S. business cycles: an empirical investigation. In: Discussion papers 451, Northwestern University, Center for Mathematical Studies in Economics and Management Science, May
• Liu C-L (2010) A tutorial of the wavelet transform. Duke University Community, 01
• LocalSolver (2021) Python api reference,
• Mendez-Jimenez I, Cardenas-Montes M (2018) Time series decomposition for improving the forecasting performance of convolutional neural networks. In: 18th Conference of the Spanish association for artificial intelligence. Springer International Publishing, pp 87–97
• Monsell B, Lytras D, Findley D (2013) Getting started with X-13ARIMA-SEATS input files (accessible version). United States Census Bureau, 08
• Montgomery DC, Jennings CL, Kulahci M (2015) Introduction to time series analysis and forecasting. John Wiley & Sons, London
  Google Scholar
• Ollech D (2018) Seasonal adjustment of daily time series. In: Bundesbank discussion paper 41/2018, Deutsche Bundesbank, Frankfurt a. M
• Persons WM (1919) Indices of business conditions. Rev Econ Stat, vol 1
• Qin L, Li W, Li S (2019) Effective passenger flow forecasting using STL and ESN based on two improvement strategies. Neurocomputing 356:05
  Article Google Scholar
• Terrell C (2019) Predictions in time series using regression models. EDTECH
• Theodosiou M (2011) Forecasting monthly and quarterly time series using STL decomposition. Int J Forecast 27:1178–1195
  Article Google Scholar
• Walker WE (1976) A heuristic adjacent extreme point algorithm for the fixed charge problem. Manag Sci 22(5):587–596
  Article Google Scholar
• Wen Q, Gao J, Song X, Sun L, Xu H, Zhu S (2019) RobustSTL: A robust seasonal-trend decomposition algorithm for long time series. In: The Thirty-Third AAAI Conference on Artificial Intelligence. AAAI Press, pp 5409–5416
• Yaffee RA, McGee M (2000) An introduction to time series analysis and forecasting with applications of SAS and SPSS. Elsevier, London
  Google Scholar
• Zhigljavsky A (2011) Singular spectrum analysis for time series. Springer, Berlin, pp 1335–1337

Download references

Acknowledgements

The second author would like to thank Osman Aydemir Yetkin and Nedim Yılmaz, from Bıçakcılar Medical Devices, for helpful feedback on this work’s earlier numerical experiments’ forecasting performance. The authors also thank two anonymous referees and guest editors for their constructive comments and suggestions.

Author information

Authors and Affiliations

1. Department of Industrial Engineering, Ozyegin University, Istanbul, Turkey
   Seyma Gozuyilmaz & O. Erhun Kundakcioglu

Authors

1. Seyma Gozuyilmaz
2. O. Erhun Kundakcioglu

Corresponding author

Correspondence toO. Erhun Kundakcioglu.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

About this article

Cite this article

Gozuyilmaz, S., Kundakcioglu, O.E. Mathematical optimization for time series decomposition.OR Spectrum 43, 733–758 (2021). https://doi.org/10.1007/s00291-021-00637-w

Download citation

• Received: 24 December 2019
• Accepted: 13 May 2021
• Published: 08 June 2021
• Version of record: 08 June 2021
• Issue date: September 2021
• DOI: https://doi.org/10.1007/s00291-021-00637-w

Keywords