We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.NA

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Numerical Analysis

Title: Time Series Forecasting Using Manifold Learning

Abstract: We address a three-tier numerical framework based on manifold learning for the forecasting of high-dimensional time series. At the first step, we embed the time series into a reduced low-dimensional space using a nonlinear manifold learning algorithm such as Locally Linear Embedding and Diffusion Maps. At the second step, we construct reduced-order regression models on the manifold, in particular Multivariate Autoregressive (MVAR) and Gaussian Process Regression (GPR) models, to forecast the embedded dynamics. At the final step, we lift the embedded time series back to the original high-dimensional space using Radial Basis Functions interpolation and Geometric Harmonics. For our illustrations, we test the forecasting performance of the proposed numerical scheme with four sets of time series: three synthetic stochastic ones resembling EEG signals produced from linear and nonlinear stochastic models with different model orders, and one real-world data set containing daily time series of 10 key foreign exchange rates (FOREX) spanning the time period 03/09/2001-29/10/2020. The forecasting performance of the proposed numerical scheme is assessed using the combinations of manifold learning, modelling and lifting approaches. We also provide a comparison with the Principal Component Analysis algorithm as well as with the naive random walk model and the MVAR and GPR models trained and implemented directly in the high-dimensional space.
Subjects: Numerical Analysis (math.NA); Machine Learning (cs.LG); Dynamical Systems (math.DS)
MSC classes: 37Mxx, 58D30, 60G15, 37C05
Cite as: arXiv:2110.03625 [math.NA]
  (or arXiv:2110.03625v4 [math.NA] for this version)

Submission history

From: Constantinos Siettos [view email]
[v1] Thu, 7 Oct 2021 17:09:59 GMT (89kb,D)
[v2] Fri, 8 Oct 2021 14:26:29 GMT (89kb,D)
[v3] Thu, 21 Oct 2021 10:29:15 GMT (91kb,D)
[v4] Tue, 23 Nov 2021 18:45:30 GMT (90kb,D)

Link back to: arXiv, form interface, contact.