Poisson-Weighted Estimation by Discrete Kernel with Application to Radiation Biodosimetry

Abstract

Reminding the framework of discrete smoothing using discrete associated kernel methods, binomial kernel with local Bayesian bandwidth selection is presented, for estimating a probability mass function under a Poisson-weighted assumption (Senga-Kiesse et al. Comput Stat 31:189–206, 2016, [11]). Model diagnostics are evoked between three approaches: parametric, nonparametric and semiparametric. Finally, some applications are done on real count datasets of low and high radiation doses in biodosimetry, as alternatives to the parametric approaches in Pujol et al. (PLoS ONE 9(12):e114137, 2014, [9]).