Novel method detects structural breaks in functional data distribution, suggesting improved analysis techniques.
Key Points
This research aims to introduce a method for detecting structural breaks in the distribution of functional data and to provide theoretical underpinnings for it.
Development of a novel kernel-based method for detecting structural breaks.
Asymptotic null distribution of the test statistic established with fewer assumptions.
Proposed unified bootstrap procedure for constructing confidence intervals for the break date.
Comprehensive simulation studies confirm the validity of the proposed method.
Application to Australian temperature data reveals significant structural breaks.
Canadian weather data shows improved goodness of fit using the proposed method.