A non-parametric estimator for the doubly-periodic Poisson intensity function
In a series of papers, J. Garrido and Y. Lu have proposed and investigated a doubly-periodic Poisson model, and then applied it to analyze hurricane data. The authors have suggested several parametric models for the underlying intensity function. In the present paper we construct and analyze a non-parametric estimator for the doubly-periodic intensity function. Assuming that only a single realization of the process is available in a bounded window, we show that the estimator is consistent and asymptotically normal when the window expands indefinitely. In addition we calculate the asymptotic bias and variance of the estimator, and in this way gain helpful information for optimizing the performance of the estimator.
|Poisson process, doubly-periodic Poisson process, periodic intensity function, non-parametric estimation, consistency, asymptotic normality, bias, variance, mean-squared error"|
|Non-Markovian processes: estimation (msc 62M09), Estimation (msc 62G05)|
|CWI. Probability, Networks and Algorithms [PNA]|
Helmers, R, Mangku, I.W, & Zitikis, R. (2007). A non-parametric estimator for the doubly-periodic Poisson intensity function. CWI. Probability, Networks and Algorithms [PNA]. CWI.