Title

Modeling and Forecasting by Manifold Learning

Department

Statistics & Analytical Sciences

Document Type

Book Chapter

Publication Date

2010

Abstract

Forecasting can benefit from recently emerged manifold learning algorithms. We describe the connection between manifold learning and forecasting. Moreover, we study the properties of different manifold learning methods and their reconstruction. These will be the technical core of the forecasting. Because manifold-based forecasting is model-free and totally nonlinear, it is promising in dealing with complex data.