Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An experimental evaluation of choices of SSA forecasting parameters

Teodor Knapik, Adolphe Ratiarison, Hasina Razafindralambo

Abstract

Six time series related to atmospheric phenomena are used as inputs for experiments offorecasting with singular spectrum analysis (SSA). Existing methods for SSA parametersselection are compared throughout their forecasting accuracy relatively to an optimal aposteriori selection and to a naive forecasting methods. The comparison shows that awidespread practice of selecting longer windows leads often to poorer predictions. It alsoconfirms that the choices of the window length and of the grouping are essential. Withthe mean error of rainfall forecasting below 1.5%, SSA appears as a viable alternative forhorizons beyond two weeks.

An experimental evaluation of choices of SSA forecasting parameters

Abstract

Six time series related to atmospheric phenomena are used as inputs for experiments offorecasting with singular spectrum analysis (SSA). Existing methods for SSA parametersselection are compared throughout their forecasting accuracy relatively to an optimal aposteriori selection and to a naive forecasting methods. The comparison shows that awidespread practice of selecting longer windows leads often to poorer predictions. It alsoconfirms that the choices of the window length and of the grouping are essential. Withthe mean error of rainfall forecasting below 1.5%, SSA appears as a viable alternative forhorizons beyond two weeks.
Paper Structure (16 sections, 19 equations, 7 figures, 2 tables)

This paper contains 16 sections, 19 equations, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. SSA steps towards the vector forecasting
  3. Embedding
  4. Singular value decomposition
  5. Obtaining elementary matrices
  6. Inverse embedding
  7. Grouping
  8. Vector forecasting
  9. Choice of the window length
  10. Choice of the grouping
  11. Data sets
  12. Experiments
  13. Results and discussion
  14. Conclusion
  15. Maximum error plots for $L^{[15]}$
  16. ...and 1 more sections

Figures (7)

  • Figure 1: Forecast mean error for all prefix groupings from [89] down to [1]
  • Figure 2: Forecast mean relative error for Maevatanana maximum temperature
  • Figure 3: Forecast mean relative error for Ambatolampy minimum temperature
  • Figure 4: Forecast mean relative error for Marovoay rainfall
  • Figure 5: Forecast mean relative error for Ambovombe vapor
  • ...and 2 more figures