Nonlinear Modeling and Prediction for Time Series
Jie Ding · Digital Access to Scholarship at Harvard (DASH) (Harvard University) · 2017
In spite of substantial results in time series analysis, there remain many unsolved problems and challenges in design of generally applicable prediction systems. In this dissertation, we address some of the challenges. We present some new techniques and analysis toward optimal model selection in well/mis-specified model classes, modeling of nonlinearity, high dimensionality, structure changes, recurring patterns, etc.