CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: Principal differential analysis: A review with extensions Authors:  Edward Gunning - University of Limerick (Ireland) [presenting]
Giles Hooker - University of California Berkeley (United States)
Abstract: Principal Differential Analysis (PDA) is a technique used to estimate time-varying linear ordinary differential equations (ODEs) from functional data. We review PDA and investigate some extensions. First, we extend the PDA model to include stochasticity in the form of smooth random disturbances to the ODE and examine the implications of this additional source of variation. Second, we investigate whether PDA can be used to approximate nonlinear time-invariant ODE models.