Skip to content

Asymptotics for Growing Dimensions

Home / Open Questions / Asymptotics for Growing Dimensions

Background: Statistical inference for multivariate forecasting evaluation often assumes a fixed number of evaluation points, but modern applications frequently involve dimensions that scale with or exceed the available sample size.

Question / Future Work: Theoretical frameworks for simultaneous inference in forecasting typically assume the dimension of the evaluation target remains constant as sample size increases. Research is needed to develop asymptotic theory that accounts for regimes where the dimensionality of the forecast evaluation space grows alongside the sample size.

Why It Matters: Addressing high-dimensional asymptotics is essential for validating the reliability of forecast comparisons in large-scale spatio-temporal and multivariate domains.

Evidence: Extending the asymptotic theory to allow J → ∞ with N → ∞ would be of interest for applications involving large spatial grids

Metadata & Links