Advertisement

Forecasting via Fokker–Planck Using Conditional Probabilities

  • Chris MontagnonEmail author
Conference paper
  • 1.1k Downloads
Part of the Contributions to Statistics book series (CONTRIB.STAT.)

Abstract

Using a closed solution to a Fokker–Planck equation model of a time series, a probability distribution for the next observation is developed. This pdf has one free parameter, b. Various approaches to selecting this parameter have been explored: most recent value, weighted moving average, etc. Here, we explore using a conditional probability distribution for this parameter b, based upon the most recent observation. These methods are tested against some real-world product sales for both a one-step ahead and a two-step ahead forecast. Significant reduction in safety stock levels is found versus an ARMA approach, without a significant increase in out-of-stocks.

Keywords

Forecasting Fokker-Planck Parameter distribution Conditional probability Stock control 

References

  1. 1.
    Kantz, H., Schreiber, T.: Analysis, Nonlinear Time Series. Cambridge University Press (2003)Google Scholar
  2. 2.
    Risken, H.: The Fokker Planck Equation. Springer (1996)Google Scholar
  3. 3.
    Noble, P.L., Wheatland, M.S.: Modeling the sunspot number distribution with a Fokker Planck equation. Astrophys. J. (2011)Google Scholar
  4. 4.
    Medino, A.V., et al.: Generalized Langevin equation driven by Levy processes: a probabilistic, numerical and time series based approach. Physica A 391 (2012)CrossRefGoogle Scholar
  5. 5.
    Czechowski, Z., Telesca, L.: Construction of a Langevin model from time series with a periodical correlation function: application to wind speed data. Physica A 392 (2013)CrossRefGoogle Scholar
  6. 6.
    Sura, P., Barsugli, J.: A note on drift and diffusion parameters from time series. Phys. Lett. A 305 (2002)Google Scholar
  7. 7.
    Kantz, H., Olbrich, E.: Deterministic and probabilistic forecasting in reconstructed spacesGoogle Scholar
  8. 8.
    Gottschall, J., Peinke, J.: On the definition and handling of different drift and diffusion estimates. New J. Phys. 10 (2008)CrossRefGoogle Scholar
  9. 9.
    Montagnon, C.: A closed solution to the Fokker-Planck equation applied to forecasting. Physica A (2014)Google Scholar
  10. 10.
    Montagnon, C.: Forecasting with the Fokker-Planck equation: Bayesian setting of parameters. Physica A (2017)Google Scholar
  11. 11.
    Montagnon, C.: Singular value decomposition and time series forecasting. Ph.D. Imperial College London (2011)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2018

Authors and Affiliations

  1. 1.Department of MathematicsImperial CollegeLondonUK

Personalised recommendations