Invariance properties and statistical inference for circular data
Gianluca Mastrantonio, Giovanna Jona Lasinio, Antonello Maruotti, Gianfranco Calise · Statistica Sinica · 2017
Statistical inference on the circle may strongly depend on the chosen reference system. Here, we introduce necessary and sucient conditions to avoid inferential problems and misinterpretation of parameter estimates for any circular distribution. The construction of invariant distributions, with respect to the reference system, is discussed by introducing specic properties. Numerical examples on articial and real data are presented to corroborate and illustrate theoretical results.